records/track_2_medium/2025-12-31_BulkSmallTrackTransfer/354be270-7d41-44b7-8064-f040923f024f.txt

import os
import sys

with open(sys.argv[0]) as f:
    code = f.read()  # read the code of this file ASAP, for logging
import copy
import glob
import math
import threading
import time
import uuid
from dataclasses import dataclass
from collections import defaultdict
from itertools import accumulate
from pathlib import Path
import gc

os.environ["PYTORCH_ALLOC_CONF"] = "expandable_segments:True"
import torch

torch.empty(
    1, device=f"cuda:{os.environ['LOCAL_RANK']}", requires_grad=True
).backward()  # prevents a bug on some systems
import torch._dynamo as dynamo
import torch.distributed as dist
import torch.nn.functional as F

# torch._inductor.config.coordinate_descent_tuning = True # we have banned this flag for new records because it causes compilation to take 30min
import triton
import triton.language as tl
from kernels import get_kernel
from torch import Tensor, nn

dynamo.config.recompile_limit = 64


# -----------------------------------------------------------------------------
# Custom operators: FP8 matmul by @YouJiacheng

@torch.library.custom_op("nanogpt::mm", mutates_args=())
def mm_op(x: Tensor, w: Tensor, x_s: float, w_s: float, grad_s: float) -> tuple[Tensor, Tensor, Tensor]:
    @torch.compile
    def impl(x: Tensor, w: Tensor):
        assert x.is_contiguous() and w.is_contiguous()
        x_f8 = x.div(x_s).to(torch.float8_e4m3fn)
        w_f8 = w.div(w_s).to(torch.float8_e4m3fn)
        out = torch._scaled_mm(
            x_f8,
            w_f8.T,
            out_dtype=torch.bfloat16,
            scale_a=x.new_tensor(x_s, dtype=torch.float32),
            scale_b=x.new_tensor(w_s, dtype=torch.float32),
            use_fast_accum=True,
        )
        return out, x_f8, w_f8

    return impl(x, w)

@mm_op.register_fake
def _(x: Tensor, w: Tensor, *_):
    assert x.ndim == w.ndim == 2
    assert x.shape[1] == w.shape[1]
    assert x.device == w.device
    assert x.is_contiguous() and w.is_contiguous()
    return x @ w.T, x.to(torch.float8_e4m3fn), w.to(torch.float8_e4m3fn)

@torch.library.custom_op("nanogpt::mm_backward", mutates_args=())
def mm_backward_op(g: Tensor, x_f8: Tensor, w_f8: Tensor, x_s: float, w_s: float, grad_s: float) -> tuple[Tensor, Tensor]:
    @torch.compile
    def impl(grad: Tensor, x_f8: Tensor, w_f8: Tensor):
        assert grad.is_contiguous()
        x_inv_s = grad.new_tensor(x_s, dtype=torch.float32)
        w_inv_s = grad.new_tensor(w_s, dtype=torch.float32)
        grad_inv_s = grad.new_tensor(grad_s, dtype=torch.float32)
        grad_f8 = grad.div(grad_s).to(torch.float8_e5m2)
        grad_x = torch._scaled_mm(
            grad_f8,
            w_f8.T.contiguous().T,
            out_dtype=torch.bfloat16,
            scale_a=grad_inv_s,
            scale_b=w_inv_s,
            use_fast_accum=False,
        )
        # faster than grad_f8_t @ x_f8, for (d_out, d_in) == (50304, 768)
        grad_w = torch._scaled_mm(
            x_f8.T.contiguous(),
            grad_f8.T.contiguous().T,
            out_dtype=torch.float32,
            scale_a=x_inv_s,
            scale_b=grad_inv_s,
            use_fast_accum=False,
        ).T
        return grad_x, grad_w

    return impl(g, x_f8, w_f8)

@mm_backward_op.register_fake
def _(g: Tensor, x_f8: Tensor, w_f8: Tensor, *_):
    return x_f8.to(torch.bfloat16), w_f8.T.contiguous().T.to(torch.float32)

def backward(ctx, grad_out: Tensor, *_):
    x_f8, w_f8 = ctx.saved_tensors
    x_s, w_s, grad_s = ctx.scales
    grad_x, grad_w = torch.ops.nanogpt.mm_backward(
        grad_out, x_f8, w_f8, x_s, w_s, grad_s
    )
    return grad_x, grad_w, None, None, None

def setup_context(ctx: torch.autograd.function.FunctionCtx, inputs, output):
    *_, x_s, w_s, grad_s = inputs
    _, x_f8, w_f8 = output
    ctx.save_for_backward(x_f8, w_f8)
    ctx.scales = x_s, w_s, grad_s
    ctx.set_materialize_grads(False)

mm_op.register_autograd(backward, setup_context=setup_context)

# -----------------------------------------------------------------------------
# Triton kernel for symmetric matrix multiplication by @byronxu99

def _get_autotune_configs():
    return [
        triton.Config(
            {
                "BLOCK_SIZE_M": bm,
                "BLOCK_SIZE_N": bn,
                "BLOCK_SIZE_K": bk,
                "GROUP_SIZE_M": 8,
                "LOWER_UPPER": 1,
            },
            num_stages=stages,
            num_warps=warps,
        )
        for bm in [64, 128]
        for bn in [64, 128, 256]
        for bk in [64, 128]
        for stages, warps in [(3, 4), (3, 8), (4, 4)]
        if bm // bn <= 2 and bn // bm <= 2
    ]

@triton.jit
def _pid_to_block(
    pid,
    M,
    BLOCK_SIZE_M: tl.constexpr,
    BLOCK_SIZE_N: tl.constexpr,
    GROUP_SIZE_M: tl.constexpr,
):
    # Split output matrix into blocks of size (BLOCK_SIZE_M, BLOCK_SIZE_N)
    num_pid_m = tl.cdiv(M, BLOCK_SIZE_M)
    num_pid_n = tl.cdiv(M, BLOCK_SIZE_N)

    # Map PID to a single matrix in batch
    batch_idx = pid // (num_pid_m * num_pid_n)
    pid = pid % (num_pid_m * num_pid_n)

    # Map PID to 2D grid of blocks
    pid_m = pid // num_pid_n
    pid_n = pid % num_pid_n
    pid_m, pid_n = tl.swizzle2d(pid_m, pid_n, num_pid_m, num_pid_n, GROUP_SIZE_M)

    m_idx = pid_m * BLOCK_SIZE_M
    n_idx = pid_n * BLOCK_SIZE_N
    return batch_idx, m_idx, n_idx

@triton.autotune(
    configs=_get_autotune_configs(),
    key=["M", "K", "a_stride_r", "a_stride_c", "c_stride_r", "c_stride_c"],
)
@triton.jit
def XXT_kernel(
    A_ptr, C_ptr,
    M, K,
    a_stride_b, a_stride_r, a_stride_c,
    c_stride_b, c_stride_r, c_stride_c,
    BLOCK_SIZE_M: tl.constexpr,
    BLOCK_SIZE_N: tl.constexpr,
    BLOCK_SIZE_K: tl.constexpr,
    GROUP_SIZE_M: tl.constexpr,
    LOWER_UPPER: tl.constexpr,
):
    pid = tl.program_id(axis=0)
    batch_idx, m_idx, n_idx = _pid_to_block(
        pid, M, BLOCK_SIZE_M, BLOCK_SIZE_N, GROUP_SIZE_M
    )

    # Skip blocks that don't need to be computed
    skip_block_below_diag = (LOWER_UPPER == 0) and (n_idx + BLOCK_SIZE_N <= m_idx)
    skip_block_above_diag = (LOWER_UPPER != 0) and (m_idx + BLOCK_SIZE_M <= n_idx)
    if skip_block_below_diag or skip_block_above_diag:
        return

    # Index into one matrix of batch
    A_ptr += batch_idx * a_stride_b
    C_ptr += batch_idx * c_stride_b

    # Create pointer arrays for A and A.T
    offs_m = (m_idx + tl.arange(0, BLOCK_SIZE_M)) % M
    offs_n = (n_idx + tl.arange(0, BLOCK_SIZE_N)) % M
    offs_k = tl.arange(0, BLOCK_SIZE_K)
    a_ptrs = A_ptr + (offs_m[:, None] * a_stride_r + offs_k[None, :] * a_stride_c)
    at_ptrs = A_ptr + (offs_k[:, None] * a_stride_c + offs_n[None, :] * a_stride_r)

    accumulator = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)

    # Accumulate over blocks of K
    for k in tl.range(0, tl.cdiv(K, BLOCK_SIZE_K)):
        a = tl.load(a_ptrs, mask=offs_k[None, :] < K - k * BLOCK_SIZE_K, other=0.0)
        at = tl.load(at_ptrs, mask=offs_k[:, None] < K - k * BLOCK_SIZE_K, other=0.0)
        accumulator = tl.dot(a, at, accumulator)
        a_ptrs += BLOCK_SIZE_K * a_stride_c
        at_ptrs += BLOCK_SIZE_K * a_stride_c

    out_dtype = C_ptr.dtype.element_ty
    output = accumulator.to(out_dtype)

    # Store block of C
    offs_cm = m_idx + tl.arange(0, BLOCK_SIZE_M)
    offs_cn = n_idx + tl.arange(0, BLOCK_SIZE_N)
    c_ptrs = C_ptr + (offs_cm[:, None] * c_stride_r + offs_cn[None, :] * c_stride_c)
    c_mask = (offs_cm[:, None] < M) & (offs_cn[None, :] < M)
    tl.store(c_ptrs, output, mask=c_mask)

    # Store block of C mirrored across the diagonal
    c_ptrs_t = C_ptr + (offs_cn[:, None] * c_stride_r + offs_cm[None, :] * c_stride_c)
    c_mask_t = (offs_cn[:, None] < M) & (offs_cm[None, :] < M)
    tl.store(c_ptrs_t, output.T, mask=c_mask_t)

def XXT(A: torch.Tensor, out: torch.Tensor):
    """
    Launch Triton kernel to compute C = A @ A.T
    """
    assert A.ndim == 2 or A.ndim == 3
    M, K = A.shape[-2:]
    assert out.size(-2) == M, "Output matrix has incorrect shape"
    assert out.size(-1) == M, "Output matrix has incorrect shape"

    batch_size = A.size(0) if A.ndim == 3 else 1
    input_batch_stride = A.stride(0) if A.ndim == 3 else 0
    output_batch_stride = out.stride(0) if out.ndim == 3 else 0

    grid = lambda meta: (
        batch_size * triton.cdiv(M, meta["BLOCK_SIZE_M"]) * triton.cdiv(M, meta["BLOCK_SIZE_N"]),
    )
    XXT_kernel[grid](
        A_ptr=A,
        C_ptr=out,
        M=M,
        K=K,
        a_stride_b=input_batch_stride,
        a_stride_r=A.stride(-2),
        a_stride_c=A.stride(-1),
        c_stride_b=output_batch_stride,
        c_stride_r=out.stride(-2),
        c_stride_c=out.stride(-1),
    )
    return out

@triton.autotune(
    configs=_get_autotune_configs(),
    key=["M", "a_stride_r", "a_stride_c", "c_stride_r", "c_stride_c"],
)
@triton.jit
def ba_plus_cAA_kernel(
    A_ptr, C_ptr,
    M,
    a_stride_b, a_stride_r, a_stride_c,
    c_stride_b, c_stride_r, c_stride_c,
    alpha, beta,
    BLOCK_SIZE_M: tl.constexpr,
    BLOCK_SIZE_N: tl.constexpr,
    BLOCK_SIZE_K: tl.constexpr,
    GROUP_SIZE_M: tl.constexpr,
    LOWER_UPPER: tl.constexpr,
):
    # This is mostly duplicated from XXT_kernel, but also loads and adds a block of A
    # Performance is slightly slower than XXT_kernel, so we use two separate kernels
    pid = tl.program_id(axis=0)
    batch_idx, m_idx, n_idx = _pid_to_block(
        pid, M, BLOCK_SIZE_M, BLOCK_SIZE_N, GROUP_SIZE_M
    )

    # Skip blocks that don't need to be computed
    skip_block_below_diag = (LOWER_UPPER == 0) and (n_idx + BLOCK_SIZE_N <= m_idx)
    skip_block_above_diag = (LOWER_UPPER != 0) and (m_idx + BLOCK_SIZE_M <= n_idx)
    if skip_block_below_diag or skip_block_above_diag:
        return

    # Index into one matrix of batch
    A_ptr += batch_idx * a_stride_b
    C_ptr += batch_idx * c_stride_b

    # Create pointer arrays for A and A.T
    offs_m = (m_idx + tl.arange(0, BLOCK_SIZE_M)) % M
    offs_n = (n_idx + tl.arange(0, BLOCK_SIZE_N)) % M
    offs_k = tl.arange(0, BLOCK_SIZE_K)
    a_ptrs = A_ptr + (offs_m[:, None] * a_stride_r + offs_k[None, :] * a_stride_c)
    at_ptrs = A_ptr + (offs_k[:, None] * a_stride_c + offs_n[None, :] * a_stride_r)

    accumulator = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)

    # Accumulate over blocks of K
    for k in tl.range(0, tl.cdiv(M, BLOCK_SIZE_K)):
        a = tl.load(a_ptrs, mask=offs_k[None, :] < M - k * BLOCK_SIZE_K, other=0.0)
        at = tl.load(at_ptrs, mask=offs_k[:, None] < M - k * BLOCK_SIZE_K, other=0.0)
        accumulator = tl.dot(a, at, accumulator)
        a_ptrs += BLOCK_SIZE_K * a_stride_c
        at_ptrs += BLOCK_SIZE_K * a_stride_c

    # Load block of A to add (corresponds to the current block of C)
    offs_am = m_idx + tl.arange(0, BLOCK_SIZE_M)
    offs_an = n_idx + tl.arange(0, BLOCK_SIZE_N)
    a_add_ptrs = A_ptr + (offs_am[:, None] * a_stride_r + offs_an[None, :] * a_stride_c)
    a_add_mask = (offs_am[:, None] < M) & (offs_an[None, :] < M)
    a_add = tl.load(a_add_ptrs, mask=a_add_mask, other=0.0).to(tl.float32)

    # Apply alpha and beta
    accumulator *= alpha
    accumulator += a_add * beta

    out_dtype = C_ptr.dtype.element_ty
    output = accumulator.to(out_dtype)

    # Store block of C
    offs_cm = m_idx + tl.arange(0, BLOCK_SIZE_M)
    offs_cn = n_idx + tl.arange(0, BLOCK_SIZE_N)
    c_ptrs = C_ptr + (offs_cm[:, None] * c_stride_r + offs_cn[None, :] * c_stride_c)
    c_mask = (offs_cm[:, None] < M) & (offs_cn[None, :] < M)
    tl.store(c_ptrs, output, mask=c_mask)

    # Store block of C mirrored across the diagonal
    c_ptrs_t = C_ptr + (offs_cn[:, None] * c_stride_r + offs_cm[None, :] * c_stride_c)
    c_mask_t = (offs_cn[:, None] < M) & (offs_cm[None, :] < M)
    tl.store(c_ptrs_t, output.T, mask=c_mask_t)

def ba_plus_cAA(A: torch.Tensor, alpha: float, beta: float, out: torch.Tensor):
    """
    Launch Triton kernel to compute C = alpha * A @ A.T + beta * A
    """
    assert A.ndim == 2 or A.ndim == 3
    M, K = A.shape[-2:]
    assert M == K, "Input matrix must be square"
    assert out.size(-2) == M
    assert out.size(-1) == M

    batch_size = A.size(0) if A.ndim == 3 else 1
    input_batch_stride = A.stride(0) if A.ndim == 3 else 0
    output_batch_stride = out.stride(0) if out.ndim == 3 else 0

    grid = lambda meta: (
        batch_size * triton.cdiv(M, meta["BLOCK_SIZE_M"]) * triton.cdiv(M, meta["BLOCK_SIZE_N"]),
    )
    ba_plus_cAA_kernel[grid](
        A_ptr=A,
        C_ptr=out,
        M=M,
        a_stride_b=input_batch_stride,
        a_stride_r=A.stride(-2),
        a_stride_c=A.stride(-1),
        c_stride_b=output_batch_stride,
        c_stride_r=out.stride(-2),
        c_stride_c=out.stride(-1),
        alpha=alpha,
        beta=beta,
    )
    return out

# Computed for num_iters=5, safety_factor=2e-2, cushion=2
polar_express_coeffs = [
    (8.156554524902461, -22.48329292557795, 15.878769915207462),
    (4.042929935166739, -2.808917465908714, 0.5000178451051316),
    (3.8916678022926607, -2.772484153217685, 0.5060648178503393),
    (3.285753657755655, -2.3681294933425376, 0.46449024233003106),
    (2.3465413258596377, -1.7097828382687081, 0.42323551169305323)
]

@torch.compile(dynamic=False, fullgraph=True) # Must use dynamic=False or else it's much slower
def polar_express(G: torch.Tensor, split_baddbmm: bool = False):
    """
    Polar Express Sign Method: https://arxiv.org/pdf/2505.16932
    by Noah Amsel, David Persson, Christopher Musco, Robert M. Gower.
    """
    X = G.bfloat16()
    if G.size(-2) > G.size(-1):
        X = X.mT

    # Ensure spectral norm is at most 1
    X = X / (X.norm(dim=(-2, -1), keepdim=True) * (1 + 2e-2) + 1e-6)

    # Allocate buffers
    X = X.contiguous()
    A = torch.empty((*X.shape[:-1], X.size(-2)), device=X.device, dtype=X.dtype)
    B = torch.empty_like(A)
    C = torch.empty_like(X)

    # Select batched vs unbatched
    if split_baddbmm:
        BX_matmul = torch.bmm if X.ndim > 2 else torch.mm
    else:
        aX_plus_BX = torch.baddbmm if X.ndim > 2 else torch.addmm

    # Perform the iterations
    for a, b, c in polar_express_coeffs:
        XXT(X, out=A)  # A = X @ X.mT
        ba_plus_cAA(A, alpha=c, beta=b, out=B)  # B = b * A + c * A @ A

        # Referencing X twice causes pytorch to make a defensive copy,
        # resulting in a cudaMemcpyAsync in baddbmm.
        # For large matrices (i.e., the mlp weights), it's faster to split
        # the operation into two kernels to avoid this.
        if split_baddbmm:
            BX_matmul(B, X, out=C)  # C = B @ X
            C.add_(X, alpha=a)      # C = C + a*X  (in-place, X only read)
        else:
            aX_plus_BX(X, B, X, beta=a, out=C)  # C = a * X + B @ X

        X, C = C, X  # Swap references to avoid unnecessary copies

    if G.size(-2) > G.size(-1):
        X = X.mT
    return X


# -----------------------------------------------------------------------------
# Compiled helpers for NorMuon by @chrisjmccormick

@torch.compile(dynamic=False, fullgraph=True)
def cautious_wd_and_update_inplace(p, v, wd_tensor, lr_tensor):
    """Cautious weight decay + parameter update. wd_tensor and lr_tensor are 0-D CPU tensors."""
    mask = (v * p) >= 0
    wd_factor = wd_tensor.to(p.dtype)
    lr_factor = lr_tensor.to(p.dtype)
    p.copy_(p - (p * mask * wd_factor * lr_factor) - (v * lr_factor))


@torch.compile(dynamic=False, fullgraph=True)
def apply_normuon_variance_reduction(v_chunk, second_momentum_buffer, beta2, red_dim):
    """NorMuon variance reduction. Algebraically fuses the normalization steps to minimize memory ops."""
    v_mean = v_chunk.float().square().mean(dim=red_dim, keepdim=True)
    red_dim_size = v_chunk.size(red_dim)
    v_norm_sq = v_mean.sum(dim=(-2, -1), keepdim=True).mul_(red_dim_size)
    v_norm = v_norm_sq.sqrt_()
    second_momentum_buffer.lerp_(v_mean.to(dtype=second_momentum_buffer.dtype), 1 - beta2)
    step_size = second_momentum_buffer.clamp_min(1e-10).rsqrt_()
    scaled_sq_sum = (v_mean * red_dim_size) * step_size.float().square()
    v_norm_new = scaled_sq_sum.sum(dim=(-2, -1), keepdim=True).sqrt_()
    final_scale = step_size * (v_norm / v_norm_new.clamp_min_(1e-10))
    return v_chunk.mul_(final_scale.type_as(v_chunk))


# -----------------------------------------------------------------------------
# NorMuon optimizer

class NorMuon(torch.optim.Optimizer):
    """
    Muon - MomentUm Orthogonalized by Newton-schulz

    https://kellerjordan.github.io/posts/muon/

    Muon internally runs standard SGD-momentum, and then performs an orthogonalization post-
    processing step, in which each 2D parameter's update is replaced with the nearest orthogonal
    matrix. To efficiently orthogonalize each update, we use a Newton-Schulz iteration, which has
    the advantage that it can be stably run in bfloat16 on the GPU.

    Warning: This optimizer should not be used for the embedding layer, the final fully connected layer,
    or any {0,1}-D parameters; those should all be optimized by a standard method (e.g., AdamW).

    Differences from standard Muon:
    - Newton-Shulz is replaced with Polar Express for the orthogonalization step
    - NorMuon adds a low-rank variance estimator similar to Adafactor. https://arxiv.org/pdf/2510.05491
    - small 1D parameters handled here instead of in Adam
    - Cautious weight decay, a gated version of decoupled weight decay
    - Custom distributed sizing:
    The model stores all attn and mlp weights in the same shape, and then updates the view as
    needed on the forward pass. This enables attn and mlp weights to be contained within the same
    dist.reduce_scatter_tensor() call. The model architecture has been customized to enable
    (n_attn_layers+n_mlp_layers*2)%8==0 for batching across 8 GPUs with zero padding on mlp and attn.
    The scheduling is:
        1. reduce scatter attn_gate (10 params 6 padding params)
        2. reduce scatter attn/mlp round 1 (10 attn params 6 mlp params)
        3. reduce scatter attn/mlp round 2 (16 mlp params)
        4. wait on step 1, then compute update of 1 and schedule all gather
        5. wait on step 2, then compute update of 2 and schedule all gather
        6. wait on step 3, then compute update of 3 and schedule all gather
            GPUs receive [2 ATTN, 2 ATTN, 2 ATTN, 2 ATTN, 2 ATTN, 2 MLP, 2 MLP, 2 MLP]
            GPUs that receive params of type attn reshape before computing update
        7. wait on 4, then compute update of 4 and schedule all gather
        8. wait for each all gather to complete and update params
    Empirically, leading with small params provides an additional 0.2s improvement.
    """
    def __init__(self, params, lr=0.02, weight_decay=0.01, momentum=0.95, beta2=0.95, custom_sizing=True):
        defaults = dict(lr=lr, weight_decay=weight_decay, momentum=momentum, beta2=beta2)
        self.world_size = dist.get_world_size() if dist.is_initialized() else 1
        # custom sizing requires 8 GPUs
        if custom_sizing and dist.get_world_size()==8:
            param_groups = self.generate_custom_param_groups(params)
        else:
            param_groups = self.generate_standard_param_groups(params)
        super().__init__(param_groups, defaults)

    def reset(self):
        # expose a reset for clearing buffers
        for group in self.param_groups:
            if "momentum_buffer" in group:
                group["momentum_buffer"].zero_()
                group["second_momentum_buffer"].zero_()

    def generate_standard_param_groups(self, params):
        """
        Use this method if running on less than 8 GPU or experimenting with additional attn or mlp modules.
        Creates one param group per module.
        """
        groups = defaultdict(list)
        for param in params:
            groups[param.label].append(param)

        param_groups = []
        for module_name, group_params in groups.items():
            chunk_size = (len(group_params) + self.world_size - 1) // self.world_size
            param_groups.append(dict(params=group_params, chunk_size=chunk_size))

        return param_groups

    def generate_custom_param_groups(self, params):
        """
        Implementation requires that a single GPU does not receive both attn
        and mlp params when a param group is split across GPUs.
        """
        params_list = list(params)
        module_group_order = ['attn_gate', 'value_embed_gate', 'attn', 'mlp'] # 16, 10, 16, 32
        group_sizes = [16, 10, 16, 16, 16]
        # 72... its actually 74
        params_list.sort(key=lambda x: module_group_order.index(x.label))
        print0(len(params_list), console=True)
        idx = 0
        assert len(params_list) == sum(group_sizes)
        param_groups = []
        for size in group_sizes:
            chunk_size = (size + self.world_size - 1) // self.world_size
            group_params = params_list[idx: idx + size]
            param_groups.append(dict(params=group_params, chunk_size=chunk_size))
            idx += size

        return param_groups

    @torch.no_grad()
    def step(self):
        # Efficient distributed step by @YouJiacheng, @KonstantinWilleke, @alexrgilbert,
        # @adricarda, @tuttyfrutyee, @vdlad, @ryanyang0, @vagrawal, @varunneal, @chrisjmccormick
        rank = dist.get_rank()
        group_infos = []
        for group in self.param_groups:
            params: list[Tensor] = group["params"]
            if not params:
                continue

            chunk_size = group["chunk_size"]
            padded_num_params = chunk_size * self.world_size

            stacked_grads = torch.empty(
                (padded_num_params, *params[0].shape),
                dtype=params[0].dtype,
                device=params[0].device
            )
            for i, p in enumerate(params):
                stacked_grads[i].copy_(p.grad, non_blocking=True)
            if len(params) < padded_num_params:
                stacked_grads[len(params):].zero_()

            grad_chunk = torch.empty_like(stacked_grads[:chunk_size])

            reduce_future = dist.reduce_scatter_tensor(
                grad_chunk, stacked_grads, op=dist.ReduceOp.AVG, async_op=True
            ).get_future()

            group_infos.append(dict(grad_chunk=grad_chunk, reduce_future=reduce_future))

        all_gather_infos = []
        # Second pass: wait for gradients, compute updates for the local shard of parameters,
        # and launch all async all_gather operations.
        for group, info in zip(self.param_groups, group_infos):
            info["reduce_future"].wait()

            params = group["params"]
            grad_chunk = info["grad_chunk"]
            chunk_size = group["chunk_size"]
            padded_num_params = chunk_size * self.world_size

            start_idx = rank * chunk_size
            module_idx = start_idx if start_idx < len(params) else 0

            num_params = min(chunk_size, max(0, len(params) - start_idx))  # num params for this rank

            if "momentum_buffer" not in group:
                group["momentum_buffer"]  = torch.zeros_like(grad_chunk[:num_params])
            momentum_buffer = group["momentum_buffer"]
            # Apply momentum update to the persistent momentum buffer in-place
            momentum_buffer.lerp_(grad_chunk[:num_params], 1 - group["momentum"])
            updated_grads = grad_chunk[:num_params].lerp_(momentum_buffer, group["momentum"])

            grad_shape = updated_grads.shape
            if params[module_idx].label == 'attn':

                for p in params[module_idx:module_idx + num_params]:
                    assert p.label == 'attn'

                updated_grads = updated_grads.view(4 * grad_shape[0], grad_shape[1] // 4, grad_shape[2])

            ref_param = params[module_idx]
            param_shape = ref_param.shape

            # The below shape-based heuristic assumes that matrices have their input along the
            # row dimension and their output along the columns. Gates are an exception.
            is_gate = 'gate' in ref_param.label

            if "second_momentum_buffer" not in group:
                if is_gate:
                    group["second_momentum_buffer"] = torch.zeros_like(updated_grads[..., :, :1])
                else:
                    group["second_momentum_buffer"] = (torch.zeros_like(updated_grads[..., :, :1])
                        if param_shape[-2] >= param_shape[-1] else torch.zeros_like(updated_grads[..., :1, :])
                    )
            second_momentum_buffer = group["second_momentum_buffer"]

            if "param_lr_cpu" not in group:
                # Define multipliers for ALL params in this group (global, not per-shard)
                lr_mults = []
                wd_mults = []
                for p in params:
                    # Increase learning rate for modules with larger inputs than outputs.
                    # This shape check also assumes rows=input, columns=output, so take care
                    # when changing memory layouts. @chrisjmccormick
                    shape = p.shape
                    if len(shape) >= 2:
                        shape_mult = max(1.0, shape[-2] / shape[-1]) ** 0.5
                    else:
                        shape_mult = 1.0
                    lr_mults.append(shape_mult * getattr(p, "lr_mul", 1.0))
                    wd_mults.append(getattr(p, "wd_mul", 1.0))
                # Define as cpu tensors to enable Inductor constant folding
                group["param_lr_cpu"] = torch.tensor(lr_mults, dtype=torch.float32, device="cpu")
                group["param_wd_cpu"] = torch.tensor(wd_mults, dtype=torch.float32, device="cpu")

            eff_lr_all = group["param_lr_cpu"] * group["lr"]
            eff_wd_all = group["param_wd_cpu"] * group["weight_decay"] * group["lr"]

            # Slice the portion corresponding to this rank's shard
            eff_lr_cpu = eff_lr_all[module_idx:module_idx + num_params]
            eff_wd_cpu = eff_wd_all[module_idx:module_idx + num_params]

            # Compute zeropower for the entire chunk in a single, batched call.
            if num_params == 0:
                v_chunk = updated_grads
            else:
                v_chunk = polar_express(updated_grads, split_baddbmm=(ref_param.label == 'mlp'))

            # Note that the head orientation in O is transposed relative to QKV, so red_dim
            # is 'incorrect' for O. However, correcting this showed no improvement. @chrisjmccormick
            red_dim = -1 if (is_gate or param_shape[-2] >= param_shape[-1]) else -2

            v_chunk = apply_normuon_variance_reduction(
                v_chunk, second_momentum_buffer, group["beta2"], red_dim
            )

            v_chunk = v_chunk.view(grad_shape)

            # # "Cautious" weight decay (https://arxiv.org/abs/2510.12402)
            updated_params = torch.empty_like(grad_chunk)
            if num_params > 0:
                # Work on a stacked copy to avoid touching original params
                param_chunk = torch.stack(params[module_idx:module_idx + num_params])

                for local_idx in range(num_params):
                    cautious_wd_and_update_inplace(
                        param_chunk[local_idx],
                        v_chunk[local_idx],
                        eff_wd_cpu[local_idx],
                        eff_lr_cpu[local_idx],
                    )
            else:
                param_chunk = torch.zeros_like(v_chunk)

            updated_params[:num_params].copy_(param_chunk)
            if num_params < chunk_size:
                updated_params[num_params:].zero_()

            stacked_params = torch.empty(
                (padded_num_params, *param_shape),
                dtype=updated_params.dtype,
                device=updated_params.device,
            )

            gather_future = dist.all_gather_into_tensor(
                stacked_params, updated_params, async_op=True
            ).get_future()

            all_gather_infos.append(
                {
                    "gather_future": gather_future,
                    "stacked_params": stacked_params,
                    "orig_params": params,
                }
            )

        # Final pass: wait for all_gather to complete and copy results back into original parameter tensors.
        for info in all_gather_infos:
            info["gather_future"].wait()
            stacked_params = info["stacked_params"]
            orig_params = info["orig_params"]

            unstacked_params = torch.unbind(stacked_params)
            for i, p in enumerate(orig_params):
                p.copy_(unstacked_params[i], non_blocking=True)


class DistAdam(torch.optim.Optimizer):
    def __init__(self, params, label_order: list[str],lr: float = 1e-3, betas: tuple[float, float] = (0.9, 0.999), eps: float = 1e-8, weight_decay: float = 0.01):
        self.world_size = dist.get_world_size() if dist.is_initialized() else 1
        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
        params = list(params)
        # Group by label, with explicit ordering for execution control.
        params_by_label = defaultdict(list)
        for p in params:
            params_by_label[getattr(p, 'label', None)].append(p)
        param_groups = []
        for label in label_order:
            if label in params_by_label:
                param_groups.append(dict(params=params_by_label[label]))
        # include any unlabeled params at the end (processed last)
        if None in params_by_label:
            param_groups.append(dict(params=params_by_label[None]))
        super().__init__(param_groups, defaults)
        # init state: small params (numel < 1024) use full-sized state, others use sharded
        for p in params:
            chunk = p if p.numel() < 1024 else p[:p.size(0) // self.world_size]
            exp_avg = torch.zeros_like(chunk, dtype=torch.bfloat16, device=p.device)
            self.state[p] = dict(step=0, exp_avg=exp_avg, exp_avg_sq=torch.zeros_like(exp_avg))
        # DistributedAdam implementation by @vagrawal, @akash5474
        self.should_sync = False
        self._reduce_scatter_hooks = []
        self._reduce_scatter_futures = {}
        self.register_backward_hooks()

    def register_backward_hooks(self):
        for group in self.param_groups:
            for param in group["params"]:
                self._reduce_scatter_hooks.append(param.register_post_accumulate_grad_hook(self._sync_gradient))

    @torch.no_grad()
    def _sync_gradient(self, param):
        if not self.should_sync:
            return

        grad = param.grad
        if param.numel() < 1024:
            # Small params: use all_reduce (no scatter/gather needed)
            self._reduce_scatter_futures[param] = (
                dist.all_reduce(grad, op=dist.ReduceOp.AVG, async_op=True).get_future(),
                grad
            )
        else:
            rank_size = grad.shape[0] // self.world_size
            if grad is not None:
                grad_slice = torch.empty_like(grad[:rank_size])
                self._reduce_scatter_futures[param] = (
                    dist.reduce_scatter_tensor(grad_slice, grad, op=dist.ReduceOp.AVG, async_op=True).get_future(),
                    grad_slice
                )

    def copy_lm_to_embed(self):
        # run at 2/3 of training
        lm_head = self.param_groups[0]['params'][0]
        embed = self.param_groups[-1]['params'][0]
        lm_head_state = self.state[lm_head]
        embed_state = self.state[embed]
        embed_state['step'] = lm_head_state['step']
        embed_state['exp_avg'] = lm_head_state['exp_avg'].clone()
        embed_state['exp_avg_sq'] = lm_head_state['exp_avg_sq'].clone()
        embed.data.copy_(lm_head.data)

    @torch.compile
    @torch.no_grad()
    def step(self):
        rank = dist.get_rank()
        all_gather_futures: list[torch.Future] = []

        for group in self.param_groups:
            beta1, beta2 = group['betas']
            eps = group['eps']
            wd = group['weight_decay']
            for param in group['params']:
                if param not in self._reduce_scatter_futures:
                    continue

                fut, g_slice = self._reduce_scatter_futures[param]
                fut.wait()

                is_small = param.numel() < 1024
                if is_small:
                    # Small params: g_slice is actually full grad, p_slice is full param
                    p_slice = param
                else:
                    rank_size = param.shape[0] // self.world_size
                    p_slice = param[rank * rank_size:(rank + 1) * rank_size]

                lr = group['lr'] * getattr(param, "lr_mul", 1.0)
                state = self.state[param]

                exp_avg = state["exp_avg"]
                exp_avg_sq = state["exp_avg_sq"]
                state["step"] += 1
                t = state["step"]
                # update running averages
                exp_avg.mul_(beta1).add_(g_slice, alpha=1 - beta1)
                exp_avg_sq.mul_(beta2).addcmul_(g_slice, g_slice, value=1 - beta2)
                # bias corrections
                bias1 = 1 - beta1 ** t
                bias2 = 1 - beta2 ** t
                # compute step
                denom = exp_avg_sq.sqrt().add_(eps)
                step_size = lr * (bias2 ** 0.5 / bias1)
                update = exp_avg.div(denom).mul_(step_size)
                # cautious weight decay
                mask = (update * p_slice) > 0
                # lr as weight decay schedule
                eff_weight_decay = lr * wd * getattr(param, "wd_mul", 1.0)
                update.addcmul_(p_slice, mask, value=eff_weight_decay * lr)

                p_slice.add_(other=update, alpha=-1.0)

                if not is_small:
                    all_gather_futures.append(dist.all_gather_into_tensor(param, p_slice, async_op=True).get_future())

        self._reduce_scatter_futures.clear()
        torch.futures.collect_all(all_gather_futures).wait()

# -----------------------------------------------------------------------------
# PyTorch nn.Module definitions for the model

def norm(x: Tensor):
    return F.rms_norm(x, (x.size(-1),))

class CastedLinear(nn.Linear):
    def __init__(self, in_features: int, out_features: int, use_fp8=False, x_s=1.0, w_s=1.0, grad_s=1.0):
        super().__init__(in_features, out_features, bias=False)
        self.use_fp8 = False # use_fp8
        self.x_s = x_s
        self.w_s = w_s
        self.grad_s = grad_s

    def reset_parameters(self) -> None:
        with torch.no_grad():
            self.weight.zero_()  # @Grad62304977 and others

    def forward(self, x: Tensor):
        if self.use_fp8 and self.training:
            _x = x.flatten(0, -2)
            out: Tensor = torch.ops.nanogpt.mm(_x, self.weight, x_s=self.x_s, w_s=self.w_s, grad_s=self.grad_s)[0]
            return out.reshape(*x.shape[:-1], -1)
        else:
            return F.linear(x, self.weight.type_as(x))


# -----------------------------------------------------------------------------
# PyTorch nn.Module definitions for the model

# yarn implementation @classiclarryd
class Yarn(nn.Module):
    def __init__(self, head_dim, max_seq_len):
        super().__init__()
        self.head_dim = head_dim
        self.max_seq_len = max_seq_len
        self.reset()

    def reset(self):
        angular_freq = (1 / 1024) ** torch.linspace(0, 1, steps=self.head_dim//4, dtype=torch.float32, device=device)
        # half-truncate RoPE by @YouJiacheng (w/ base freq tuning)
        angular_freq = torch.cat([angular_freq, angular_freq.new_zeros(self.head_dim//4)])
        t = torch.arange(self.max_seq_len, dtype=torch.float32, device=device)
        theta = torch.outer(t, angular_freq)
        self.cos = nn.Buffer(
            theta.cos().to(torch.bfloat16), persistent=False
        )
        self.sin = nn.Buffer(
            theta.sin().to(torch.bfloat16), persistent=False
        )
        self.angular_freq = angular_freq
        # start with 0.1, inspired by 0.12 from @leloykun and learnable scalars used by @brendanh0gan https://x.com/hi_tysam/status/1879693583898591283
        self.attn_scale = 0.1

    def apply(self, old_window: int, new_window: int, alpha: int=1, beta: int=32):
        rotations = args.block_size * old_window * self.angular_freq / (2 * torch.pi)
        scaling_factor = old_window / new_window
        interpolation_weight = torch.clamp((rotations - alpha) / (beta - alpha), 0, 1)
        self.angular_freq *= scaling_factor + interpolation_weight * (1 - scaling_factor)
        t = torch.arange(self.max_seq_len, dtype=torch.float32, device=self.angular_freq.device)
        theta = torch.outer(t, self.angular_freq)
        self.cos.copy_(theta.cos())
        self.sin.copy_(theta.sin())
        self.attn_scale *= 0.2 * math.log(new_window / old_window) + 1

def rotary(x_BTHD: Tensor, cos: Tensor, sin: Tensor):
    assert cos.size(0) >= x_BTHD.size(-3)
    cos, sin = (
        cos[None, : x_BTHD.size(-3), None, :],
        sin[None, : x_BTHD.size(-3), None, :],
    )
    x1, x2 = x_BTHD.chunk(2, dim=-1)
    y1 = x1 * cos + x2 * sin
    y2 = x1 * (-sin) + x2 * cos
    return torch.cat((y1, y2), 3)

@dataclass
class AttnArgs:
    ve: torch.Tensor
    sa_lambdas: torch.Tensor
    seqlens: torch.Tensor
    bm_size: int
    cos: torch.Tensor
    sin: torch.Tensor
    attn_scale: float
    key_offset: bool

flash_attn_interface = get_kernel('varunneal/flash-attention-3').flash_attn_interface

class CausalSelfAttention(nn.Module):
    def __init__(self, dim: int, head_dim: int, num_heads: int, layer_idx: int):
        super().__init__()
        self.num_heads = num_heads
        self.head_dim = head_dim
        self.dim = dim
        self.hdim = num_heads * head_dim

        assert self.hdim == self.dim, "num_heads * head_dim must equal model_dim"
        std = self.dim ** -0.5
        bound = (3 ** 0.5) * std # improved init scale by @YouJiacheng
        # merged QKVO weights: suggested by many, implemented by @fernbear.bsky.social, and further improved by @YouJiacheng
        # https://x.com/hi_tysam/status/1879699187107033311
        # Simplified layout by @chrisjmccormick
        self.qkvo_w = nn.Parameter(torch.empty(self.dim * 4, self.hdim))
        # label all modules for explicit optimizer grouping
        self.qkvo_w.label = 'attn'

        with torch.no_grad():
            self.qkvo_w[:self.dim * 3].uniform_(-bound, bound)  # init QKV weights
            self.qkvo_w[self.dim * 3:].zero_()  # init O weights to zero

        # sparse gated attention to enable context based no-op by @classiclarryd
        self.attn_gate = CastedLinear(16, num_heads)
        self.attn_gate.weight.label = 'attn_gate'
        self.attn_gate.weight.lr_mul = 0.1

        # only include gates on layers with value embeds used on forward pass
        if layer_idx in [0,1,2,3,4,11, 12, 13, 14, 15]:
            self.value_embed_gate = CastedLinear(16, num_heads)
            self.value_embed_gate.weight.label = 'value_embed_gate'
            self.value_embed_gate.weight.lr_mul = 0.1

    def forward(self, x: Tensor, attn_args: AttnArgs):
        B, T = x.size(0), x.size(1) # batch size, sequence length
        assert B == 1, "varlen sequences requires B == 1"
        assert T % 16 == 0
        # unpack attention args
        cos, sin = attn_args.cos, attn_args.sin
        ve, sa_lambdas, key_offset = attn_args.ve, attn_args.sa_lambdas, attn_args.key_offset
        seqlens, attn_scale, bm_size = attn_args.seqlens, attn_args.attn_scale, attn_args.bm_size

        q, k, v = F.linear(x, sa_lambdas[0] * self.qkvo_w[:self.dim * 3].type_as(x)).view(B, T, 3 * self.num_heads, self.head_dim).chunk(3, dim=-2)
        q, k = norm(q), norm(k) # QK norm @Grad62304977
        q, k = rotary(q, cos, sin), rotary(k, cos, sin)
        if key_offset:
            # shift keys forward for the stationary head dims. Enables 1-layer induction.
            k[:, 1:, :, self.head_dim // 4:self.head_dim // 2] = k[:, :-1, :, self.head_dim // 4:self.head_dim // 2]
            k[:, 1:, :, 3 * self.head_dim // 4:] = k[:, :-1, :, 3 * self.head_dim // 4:]
        if ve is not None:
            ve_gate_out = 2 * torch.sigmoid(self.value_embed_gate(x[..., :self.value_embed_gate.weight.size(-1)])).view(B, T, self.num_heads, 1)
            v = v + ve_gate_out * ve.view_as(v) # @ KoszarskyB & @Grad62304977

        max_len = args.train_max_seq_len if self.training else (args.val_batch_size // (grad_accum_steps * world_size))

        # use flash_attn over flex_attn @varunneal. flash_attn_varlen suggested by @YouJiacheng
        y = flash_attn_interface.flash_attn_varlen_func(q[0], k[0], v[0], cu_seqlens_q=seqlens, cu_seqlens_k=seqlens,
                                                        max_seqlen_q=max_len, max_seqlen_k=max_len,
                                                        causal=True, softmax_scale=attn_scale, window_size=(bm_size, 0))
        y = y.view(B, T, self.num_heads, self.head_dim)
        y = y * torch.sigmoid(self.attn_gate(x[..., :self.attn_gate.weight.size(-1)])).view(B, T, self.num_heads, 1)
        y = y.contiguous().view(B, T, self.num_heads * self.head_dim) # re-assemble all head outputs side by side
        y = F.linear(y, sa_lambdas[1] * self.qkvo_w[self.dim * 3:].type_as(y))  # sa_lambdas[1] pre-multiplied to O @shenberg
        return y

class MLP(nn.Module):
    def __init__(self, dim: int):
        super().__init__()
        hdim = 4 * dim
        # Transposed layout to match attention weights
        self.c_fc = nn.Parameter(torch.empty(hdim, dim))
        self.c_proj = nn.Parameter(torch.empty(hdim, dim))
        # label all modules for explicit optimizer grouping
        self.c_fc.label = 'mlp'
        self.c_proj.label = 'mlp'
        self.c_proj.lr_mul = 2.

        std = 0.5 * (dim ** -0.5)
        bound = (3 ** 0.5) * std # improved init scale by @YouJiacheng
        with torch.no_grad():
            self.c_fc.uniform_(-bound, bound)
            self.c_proj.zero_() # zero init suggested by @Grad62304977

    def forward(self, x: Tensor):
        x = F.linear(x, self.c_fc.type_as(x))
        x = F.relu(x).square() # https://arxiv.org/abs/2109.08668v2; ~1-2% better than GELU; suggested by @SKYLINEZ007 and @Grad62304977
        x = F.linear(x, self.c_proj.T.type_as(x))
        return x

class Block(nn.Module):
    def __init__(self, dim: int, head_dim: int, num_heads: int, layer_idx: int):
        super().__init__()
        # skip attention of blocks.6 (the 7th layer) by @YouJiacheng
        self.attn = CausalSelfAttention(dim, head_dim, num_heads, layer_idx)
        # skip MLP blocks for first MLP layer by @EmelyanenkoK
        self.mlp = MLP(dim)

    def forward(self, x: Tensor, attn_args: AttnArgs):
        if self.attn is not None:
            x = x + self.attn(norm(x), attn_args)
        if self.mlp is not None:
            x = x + self.mlp(norm(x))
        return x

# -----------------------------------------------------------------------------
# The main model

def next_multiple_of_n(v: float | int, *, n: int):
    return next(x for x in range(n, int(v) + 1 + n, n) if x >= v)

@dataclass
class ForwardScheduleConfig:
    mtp_weights: torch.Tensor
    ws_short: int
    ws_long: int

class GPT(nn.Module):
    def __init__(self, vocab_size: int, num_layers: int, num_heads: int, head_dim: int, model_dim: int, max_seq_len: int):
        super().__init__()
        self.num_layers = num_layers
        vocab_size = next_multiple_of_n(vocab_size, n=128)

        self.smear_gate = CastedLinear(16, 1)
        self.smear_gate.weight.label = 'smear_gate'
        self.smear_gate.weight.lr_mul = 0.01
        self.smear_gate.weight.wd_mul = 0.0

        self.skip_gates = nn.ModuleList([CastedLinear(16, 1) for _ in range(3)])
        for sg in self.skip_gates:
            sg.weight.label = 'skip_gate'
            sg.weight.lr_mul = 0.01
            sg.weight.wd_mul = 0.0

        # token value embeddings by @KoszarskyB - inspired by @Grad62304977's value residual implementation following https://arxiv.org/abs/2410.17897
        # value embedding code simplification inspired by @ragulpr https://github.com/KellerJordan/modded-nanogpt/pull/78
        self.value_embeds = nn.ModuleList([nn.Embedding(vocab_size, model_dim) for _ in range(5)])
        for embed in self.value_embeds:
            nn.init.zeros_(embed.weight)
        for ve in self.value_embeds:
            ve.weight.label = 'value_embed'
        self.blocks = nn.ModuleList([Block(model_dim, head_dim, num_heads, i) for i in range(num_layers)])
        self.yarn = Yarn(head_dim, max_seq_len)
        # there are only 50257 unique GPT-2 tokens; we extend to nearest multiple of 128 for efficiency.
        # suggested to me by @Grad62304977. this originates from Karpathy's experiments.
        use_fp8 = not os.environ.get("DISABLE_FP8", False)

        self.lm_head = CastedLinear(model_dim, vocab_size, use_fp8=use_fp8, x_s=100/448, w_s=1.6/448, grad_s=0.75/448)
        nn.init.normal_(self.lm_head.weight, mean=0, std=0.005)
        self.lm_head.weight.label = 'lm_head'

        self.embed = nn.Embedding(vocab_size, model_dim)
        self.embed.weight.label = 'embed'

        self.embed2 = nn.Embedding(vocab_size, model_dim)
        self.embed2.weight.label = 'embed2'

        # x0_lambdas separated out for different optimizer treatment (no beta smoothing)
        self.x0_lambdas = nn.Parameter(torch.zeros(2*num_layers))
        self.x0_lambdas.label = 'x0_lambdas'
        self.x0_lambdas.lr_mul = 5.0
        self.x0_lambdas.wd_mul = 0.0

        pad = (-num_layers * 3 - 5) % dist.get_world_size()  # updated: 3*num_layers instead of 4*
        self.scalars = nn.Parameter(
            torch.cat(
                [
                    1.05 * torch.ones(num_layers),  # resid lambdas. 1.05 init such that layer i weight is i^(num_layers-i).
                    *[torch.tensor([0.5, 1.0]) for _ in range(num_layers)],  # SA lambdas
                    torch.zeros(1), # smear_lambda
                    0.5*torch.ones(1), # backout_lambda
                    -1.5 * torch.ones(3),  # skip_lambdas -> σ(-1.5) ≈ 0.18
                    torch.ones(pad),
                ]
            )
        )

        self.scalars.label = 'scalars'
        # set learning rates
        for param in self.value_embeds.parameters():
            param.lr_mul = 75.
            param.wd_mul = 5.
        for param in self.embed.parameters():
            param.wd_mul = 150.
        for param in self.embed2.parameters():
            param.lr_mul = 75.
            param.wd_mul = 5.
        for param in self.lm_head.parameters():
            param.wd_mul = 150.
        self.scalars.lr_mul = 5.0
        self.scalars.wd_mul = 0.0

        self.split_embed = False

    def forward(self, input_seq: Tensor, target_seq: Tensor, seqlens: Tensor, schedule_cfg: ForwardScheduleConfig):
        assert input_seq.ndim == 1

        # unpack schedule_cfg
        mtp_weights, ws_short, ws_long = schedule_cfg.mtp_weights, schedule_cfg.ws_short, schedule_cfg.ws_long

        # set configs
        skip_connections = []
        skip_in = [2, 4, 6]
        skip_out = [9, 10, 11]
        x_backout = None
        backout_layer = 11

        # set lambdas
        resid_lambdas = self.scalars[: 1 * self.num_layers]
        x0_lambdas = self.x0_lambdas.view(-1, 2)
        sa_lambdas = self.scalars[1 * self.num_layers: 3 * self.num_layers].view(-1, 2)
        smear_lambda = self.scalars[3 * self.num_layers]
        backout_lambda = self.scalars[3 * self.num_layers+1]
        skip_lambdas = self.scalars[3 * self.num_layers+2: 3*self.num_layers+5]
        
        # set block masks and key shift
        short_bm = ws_short * args.block_size
        long_bm = ws_long * args.block_size
        bm_sizes = [long_bm, short_bm, short_bm, short_bm, long_bm, short_bm, short_bm, short_bm, 
            short_bm, short_bm, short_bm, long_bm, short_bm, short_bm, short_bm, long_bm]
        assert len(bm_sizes) == self.num_layers
        key_offset = [b==long_bm for b in bm_sizes] # apply partial key offset to long windows

        # weight-tied: use lm_head.weight for embedding lookup (or separate embed after split)
        if self.split_embed:
            x = self.embed(input_seq)
        else:
            x = F.embedding(input_seq, self.lm_head.weight)
        ve = [value_embed(input_seq) for value_embed in self.value_embeds]
        # 012 ... 012 structure on token value embeddings by @YouJiacheng, improved on @leloykun's U-net structure
        # dropping first layer updates this to .12 ... 012
        ve = [ve[0], ve[1], ve[2], ve[3], ve[4]] + [None] * (self.num_layers - 10) + [ve[0], ve[1], ve[2], ve[3], ve[4]]
        assert len(ve) == self.num_layers

        # smear token embed forward 1 position @classiclarryd
        smear_gate_out = smear_lambda * torch.sigmoid(self.smear_gate(x[1:, :self.smear_gate.weight.size(-1)]))
        x = torch.cat([x[:1], x[1:] + smear_gate_out * x[:-1]])
        x = x0 = norm(x[None])

        x02 = norm(self.embed2(input_seq)[None])

        skip_idx = 0
        for i in range(self.num_layers):
            attn_args = AttnArgs(
                ve=ve[i],
                sa_lambdas=sa_lambdas[i],
                seqlens=seqlens,
                bm_size=bm_sizes[i],
                cos=self.yarn.cos,
                sin=self.yarn.sin,
                attn_scale=self.yarn.attn_scale,
                key_offset = key_offset[i]
            )
            if i in skip_out:
                skip_gate_out = torch.sigmoid(skip_lambdas[skip_idx]) * 2 * torch.sigmoid(self.skip_gates[skip_idx](x0[..., :self.skip_gates[skip_idx].weight.size(-1)]))
                skip_idx += 1
                x = x + skip_gate_out * skip_connections.pop()
            if i == 0:
                x = (resid_lambdas[0] + x0_lambdas[0,0]) * x + x0_lambdas[0,1] * x02
            else:
                x = resid_lambdas[i] * x + x0_lambdas[i,0] * x0 + x0_lambdas[i,1] * x02
            x = self.blocks[i](x, attn_args)
            if i in skip_in:
                skip_connections.append(x)
            if i == backout_layer:
                x_backout = x

        # back out contributions from first 2/3 layers that are only required for downstream context and not direct prediction
        x -= backout_lambda * x_backout
        x = norm(x)

        if not self.training:
            loss = 0
            for i in range(4):
                logits: Tensor = F.linear(x.flatten(end_dim=1).chunk(4)[i], self.lm_head.weight.bfloat16()).float()
                logits = 23 * torch.sigmoid((logits + 5) / 7.5)
                loss += F.cross_entropy(logits.view(-1, logits.size(-1)), target_seq.chunk(4)[i], reduction="mean")/4
            return loss
        
        logits = self.lm_head(x)
        # @Grad62304977 added tanh softcapping following Gemma 2 paper, @KoszarskyB reduced it from 30 to 15
        # @YouJiacheng shifted it by +15 (2*sigmoid(2*x)=tanh(x)+1). @classiclarryd updated to 23*sigmoid((logits+5)/7.5)
        logits = 23 * torch.sigmoid((logits + 5) / 7.5)
        logits_for_loss = logits.float() if not self.training else logits

        n_predict = mtp_weights.size(0) if mtp_weights is not None else 1
        if self.training and n_predict > 1:
            # Multi-token prediction: take loss of the weighted average of next n_predict tokens
            logits_flat = logits_for_loss.view(-1, logits_for_loss.size(-1))
            idx = F.pad(target_seq, (0, n_predict - 1)).unfold(0, n_predict, 1)  # [T, n_predict] of shifted targets
            target_logits = logits_flat.gather(1, idx)
            cross_entropy = torch.logsumexp(logits_flat, dim=-1).unsqueeze(1) - target_logits
            for k in range(1, n_predict):  # zero out preds past end of sequence
                cross_entropy[-k:, k] = 0
            loss = (cross_entropy * mtp_weights).sum()
        elif self.training:
            loss = F.cross_entropy(logits_for_loss.view(-1, logits_for_loss.size(-1)), target_seq, reduction="sum")
        else:
            loss = F.cross_entropy(logits_for_loss.view(-1, logits_for_loss.size(-1)), target_seq, reduction="mean")
        return loss

# -----------------------------------------------------------------------------
# Distributed data loader

def _load_data_shard(file: Path):
    header = torch.from_file(str(file), False, 256, dtype=torch.int32) # header is 256 int32
    assert header[0] == 20240520, "magic number mismatch in the data .bin file"
    assert header[1] == 1, "unsupported version"
    num_tokens = int(header[2]) # number of tokens (claimed)
    with file.open("rb", buffering=0) as f:
        tokens = torch.empty(num_tokens, dtype=torch.uint16, pin_memory=True) # avoid pin_memory copy by @YouJiacheng
        f.seek(256 * 4)
        nbytes = f.readinto(tokens.numpy()) # avoid bytes->array copy by @YouJiacheng
        assert nbytes == 2 * num_tokens, "number of tokens read does not match header"
    return tokens

BOS_ID = 50256

class BOSFinder:
    # Helper for getting sequences that start at the beginning of documents by @varunneal based on work by @classiclarryd
    def __init__(self, tokens: Tensor, world_size: int = 1, quickload: bool = False):
        # Precompute BOS positions once per shard
        self.tokens=tokens
        self.size = tokens.numel()
        self.quickload = quickload
        if quickload:
            # only scan first 4 million tokens, then kickoff async thread to scan rest
            self.bos_idx = (tokens[:4_000_000] == BOS_ID).nonzero(as_tuple=True)[0].to(torch.int64).cpu().numpy()
            self.thread = None
            self.ready = threading.Event()
            self.start()
        else:
            self.bos_idx = (tokens == BOS_ID).nonzero(as_tuple=True)[0].to(torch.int64).cpu().numpy()
        self.i = 0
        self.world_size = world_size
        self.batch_iter = 0

    def _load(self):
        self.bos_idx_async = (self.tokens == BOS_ID).nonzero(as_tuple=True)[0].to(torch.int64).cpu().numpy()
        self.ready.set()

    def start(self):
        self.ready.clear()
        self.thread = threading.Thread(target=self._load)
        self.thread.start()

    def get(self):
        if self.thread:
            self.ready.wait()
            self.thread.join()
        self.bos_idx = self.bos_idx_async

    def next_batch(self, num_tokens_local: int, max_seq_len: int):
        # if quickload was used, repoint to the full dataset after 5 batches
        if self.quickload and self.batch_iter==5:
            self.get()
        n = len(self.bos_idx)
        starts = [[] for _ in range(self.world_size)]
        ends = [[] for _ in range(self.world_size)]

        idx = self.i
        for r in range(self.world_size):
            cur_len = 0
            while cur_len <= num_tokens_local:
                if idx >= n:
                    raise StopIteration(f"Insufficient BOS ahead; hit tail of shard.")
                cur = self.bos_idx[idx]
                starts[r].append(cur)
                end = min(self.bos_idx[idx + 1] if idx + 1 < n else self.size,
                          cur + max_seq_len,
                          cur + num_tokens_local - cur_len + 1)
                ends[r].append(end)
                cur_len += end - cur
                idx += 1

            assert cur_len == num_tokens_local + 1
        self.i = idx
        self.batch_iter+=1
        return starts, ends

class DataPreloader:
    # Helper for asynchronously loading next shard and indexing bos tokens
    def __init__(self, file_iter, world_size: int = 1):
        self.file_iter = file_iter
        self.world_size = world_size
        self.thread = None
        self.data = None
        self.ready = threading.Event()

    def _load(self):
        tokens = _load_data_shard(next(self.file_iter))
        self.data = (tokens, BOSFinder(tokens, self.world_size))
        self.ready.set()

    def start(self):
        self.ready.clear()
        self.thread = threading.Thread(target=self._load)
        self.thread.start()

    def get(self):
        if self.thread:
            self.ready.wait()
            self.thread.join()
        return self.data

def distributed_data_generator(filename_pattern: str, num_tokens: int, max_seq_len: int, grad_accum_steps: int = 1, align_to_bos: bool = True):
    # align_to_bos: each sequence begins with Beginning of Sequence token, sequences truncated to max_seq_len
    rank = dist.get_rank() if dist.is_initialized() else 0
    world_size = dist.get_world_size() if dist.is_initialized() else 1
    assert num_tokens % (world_size * grad_accum_steps) == 0, "Batch size must be divisible by world size"
    num_tokens = num_tokens // grad_accum_steps

    files = [Path(file) for file in sorted(glob.glob(filename_pattern))]
    if not files:
        raise FileNotFoundError(f"No files found for pattern: {filename_pattern}")

    file_iter = iter(files)  # Use itertools.cycle(files) for multi-epoch training
    tokens = _load_data_shard(next(file_iter))
    if align_to_bos:
        finder = BOSFinder(tokens, world_size=world_size, quickload=True)
        preloader = DataPreloader(file_iter, world_size)
        preloader.start()
    else:
        pos = 0  # for unaligned case

    while True:
        num_tokens_local = num_tokens // world_size
        max_num_docs = next_multiple_of_n(num_tokens_local // 300, n=128)  # median doc length is ~400

        if align_to_bos:
            try:
                seq_starts, seq_ends = finder.next_batch(num_tokens_local, max_seq_len)
                start_idxs, end_idxs = torch.tensor(seq_starts[rank]), torch.tensor(seq_ends[rank])
            except StopIteration:
                # This shard is exhausted, load the next one in the next loop iteration.
                tokens, finder = preloader.get()
                preloader.start()
                continue

            buf = torch.cat([tokens[i:j] for i, j in zip(start_idxs, end_idxs)])
            _inputs = buf[:-1]
            _targets = buf[1:]
            end_idxs[-1] -= 1  # last document was too long to account for _targets offset
            cum_lengths = (end_idxs - start_idxs).cumsum(0)

        else:
            if pos + num_tokens + 1 >= len(tokens):  # should not occur for val data
                tokens, pos = _load_data_shard(next(file_iter)), 0

            pos_local = pos + rank * num_tokens_local
            buf = tokens[pos_local: pos_local + num_tokens_local + 1]
            _inputs = buf[:-1].view(num_tokens_local, )
            _targets = buf[1:].view(num_tokens_local, )

            cum_lengths = torch.nonzero(_inputs == BOS_ID)[:, 0]
            pos += num_tokens


        _cum_lengths = torch.full((max_num_docs,), num_tokens_local)
        _cum_lengths[0] = 0
        _cum_lengths[1:len(cum_lengths) + 1] = cum_lengths

        # Cast to int32 on CPU before transfer to avoid dtype conversion during .to()
        _inputs = _inputs.to(dtype=torch.int32)
        _targets = _targets.to(dtype=torch.int64)
        _cum_lengths = _cum_lengths.to(dtype=torch.int32)

        new_params = yield (
            _inputs.to(device="cuda", non_blocking=True),
            _targets.to(device="cuda", non_blocking=True),
            _cum_lengths.to(device="cuda", non_blocking=True)
        )

        if new_params is not None:
            # makes it possible for generator to receive new (num_tokens, max_seq_len, grad_accum_steps) via .send()
            new_num_tokens, new_max_seq_len, new_grad_accum_steps = new_params
            assert new_num_tokens % (world_size * new_grad_accum_steps) == 0, "Num tokens must be divisible by world size"
            num_tokens = new_num_tokens // new_grad_accum_steps
            max_seq_len = new_max_seq_len

# -----------------------------------------------------------------------------
# Training Management

def get_bs(step: int):
    if step >= args.num_scheduled_iterations:
        return args.train_bs_extension
    x = step / args.num_scheduled_iterations
    bs_idx = int(len(args.train_bs_schedule) * x)
    return args.train_bs_schedule[bs_idx]

def get_ws(step: int):
    # set short window size to half of long window size
    # Higher ws on "extension" steps
    if step >= args.num_scheduled_iterations:
        return args.ws_final // 2, args.ws_final
    x = step / args.num_scheduled_iterations
    assert 0 <= x < 1
    ws_idx = int(len(args.ws_schedule) * x)
    return min(11,args.ws_schedule[ws_idx] // 2), args.ws_schedule[ws_idx]

# learning rate schedule: tied to batch size schedule, with cooldown at the end.
def get_lr(step: int):
    if step > args.num_scheduled_iterations:
        return 0.1
    lr_max = 1.0
    x = step / args.num_scheduled_iterations
    if x > 1/12:
       lr_max = 1.52  # (16/8)**0.6
    if x > 2/12:
        lr_max = 1.73  # (24/8)**0.5
    if x > 3/12:
        lr_max = 2.0
    if x >= 1 - args.cooldown_frac:
        w = (1 - x) / args.cooldown_frac
        lr = lr_max * w + (1 - w) * 0.1
        return lr
    return lr_max

def get_muon_momentum(step: int, muon_warmup_steps=300, muon_cooldown_steps=50, momentum_min=0.85, momentum_max=0.95):
    # warmup phase: linearly increase momentum from min to max
    # cooldown phase: linearly decrease momentum from max to min
    momentum_cd_start = args.num_iterations - muon_cooldown_steps
    if step < muon_warmup_steps:
        frac = step / muon_warmup_steps
        momentum = momentum_min + frac * (momentum_max - momentum_min)
    elif step > momentum_cd_start:
        frac = (step - momentum_cd_start) / muon_cooldown_steps
        momentum = momentum_max - frac * (momentum_max - momentum_min)
    else:
        momentum = momentum_max
    return momentum

class TrainingManager():
    """
    Manages three optimizers for Adam embed/lm_head, Adam scalars, and Muon weight matrices.
    Notable Features:
        1. Scalars are given higher momentum terms to smooth learning @ChrisJMcCormick
        2. Scalar weights are temporarily frozen during batch size or window size updates @ChrisJMcCormick
        3. Adam optimizers are only stepped on odd steps @classiclarryd
        4. Adam optimizers have hooks to start gradient communication during backwards pass @akash5474
        5. Muon has a linear momentum warmup and cooldown schedule
        6. Learning rates follow a linear decay schedule
        7. Embed/lm_head weights and optimizer state splits at 2/3 of training @classiclarryd

    Manages model architecture, data, and target that changes during training
    Notable Features:
        1. Multi Token Prediction schedule of [1, 0.5, 0.25->0] -> [1, 0.5->0] -> [1] @varunneal
        2. Sliding Attention window schedule of [1,3] -> [3,7] -> [5,11] -> [6,13]
        3. YaRN updates to RoPE on window changes
        4. Split embed and lm head at 2/3 of training
        5. Batch size schedule of 8 -> 16 -> 24
        6. Post training extension of long windows from 13 to 20
    """
    def __init__(self, model):
        self.mtp_weights_schedule = self._build_mtp_schedule()
        self.model = model

        adam_labels = ['lm_head', 'value_embed', 'smear_gate', 'skip_gate', 'x0_lambdas', 'embed2', 'embed']
        scalar_labels = ['scalars']
        muon_labels = ['attn_gate', 'value_embed_gate', 'attn', 'mlp']
        adam_params = [p for p in model.parameters() if getattr(p, 'label', None) in adam_labels]
        scalar_params = [p for p in model.parameters() if getattr(p, 'label', None) in scalar_labels]
        muon_params = [p for p in model.parameters() if getattr(p, 'label', None) in muon_labels]
        assert set(getattr(p, 'label', None) for p in model.parameters()) == set(adam_labels + scalar_labels + muon_labels), "All params must have label"

        self.adam_opt = DistAdam(adam_params, adam_labels, lr=0.004, betas=(0.8, 0.95), eps=1e-8, weight_decay=0.005)
        self.scalar_opt = DistAdam(scalar_params, scalar_labels, lr=0.008, betas=(0.9, 0.99), eps=1e-8, weight_decay=0.005)
        self.muon_opt = NorMuon(muon_params, lr=0.015, momentum=0.95, beta2=0.95, weight_decay=1.2)
        self.optimizers = [self.adam_opt, self.scalar_opt, self.muon_opt]
        # split after odd number step
        self.split_step = math.ceil(args.split_embed_frac * args.num_scheduled_iterations) | 1

        # set defaults
        for opt in self.optimizers:
            opt.freeze_timer = 0
            opt.odd_step_only = False 
            opt.should_sync = True
            for group in opt.param_groups:
                group["initial_lr"] = group["lr"]

        # on even steps, only step Muon params
        self.adam_opt.odd_step_only = True
        self.scalar_opt.odd_step_only = True

        self.reset()

    def _build_mtp_schedule(self):
        # Precompute MTP weights for all steps to avoid tensor allocation during training
        # Schedule: [1, 0.5, 0.25->0] -> [1, 0.5->0] -> [1]
        mtp_weights_schedule = []
        for s in range(args.num_iterations + 1):
            x = s / (args.num_scheduled_iterations/4)
            if x < 1/3:
                w = [1.0, 0.5, 0.25 * (1 - 3*x)]
            elif x < 2/3:
                w = [1.0, 0.5 * (1 - (3*x - 1))]
            else:
                w = [1.0]
            mtp_weights_schedule.append(torch.tensor(w, device=device))
        return mtp_weights_schedule

    def apply_final_ws_ext(self):
        self.ws_long = args.ws_validate_post_yarn_ext

    def get_forward_args(self):
        return ForwardScheduleConfig(
            mtp_weights = self.mtp_weights,
            ws_short = self.ws_short,
            ws_long = self.ws_long
        )
    
    def _is_active_step(self, opt, step: int):
        return (opt.odd_step_only and step%2==1) or not opt.odd_step_only

    def get_transition_steps(self):
        transition_steps = [0]
        ws_short, ws_long = get_ws(0)
        for step in range(1, args.num_iterations):
            ws_short, new_ws_long = get_ws(step)
            if new_ws_long != ws_long:
                transition_steps.append(step)
                ws_long = new_ws_long
        return transition_steps

    def advance_schedule(self, step: int):
        self.ws_short, new_ws_long = get_ws(step)
        # only apply yarn for first few
        if new_ws_long != self.ws_long and new_ws_long<=13:
            self.model.yarn.apply(self.ws_long, new_ws_long)

        new_batch_size = get_bs(step)
        if new_batch_size != self.batch_size:
            self.train_loader_send_args = (new_batch_size, args.train_max_seq_len, grad_accum_steps)
        else:
            self.train_loader_send_args = None

        self.ws_long = new_ws_long
        self.mtp_weights = self.mtp_weights_schedule[step]
    
    def step_optimizers(self, step: int):                
        step_lr = get_lr(step)
        muon_momentum = get_muon_momentum(step)
        for group in self.muon_opt.param_groups:
            group["momentum"] = muon_momentum

        for opt in self.optimizers:
            if opt.freeze_timer > 0:
                opt.freeze_timer -= 1
                opt.zero_grad(set_to_none=True)
            else:
                if self._is_active_step(opt, step):
                    for group in opt.param_groups:
                        group["lr"] = group["initial_lr"] * step_lr
                    opt.step()
                    opt.zero_grad(set_to_none=True)
                    if opt.odd_step_only:
                        opt.should_sync = False
        
        if step == self.split_step:
            self.adam_opt.copy_lm_to_embed()
            self.model.split_embed = True

    def start_transition(self, freeze_count=40):
        # freeze scalar weights during transition
        self.scalar_opt.freeze_timer = freeze_count
    
    def activate_hooks(self, step: int):
        for opt in self.optimizers:
            if self._is_active_step(opt, step):
                opt.should_sync = True

    def reset(self, state=None):
        if state is not None:
            for opt, opt_state in zip(self.optimizers, state):
                opt.should_sync = False
                opt.load_state_dict(opt_state)

        # muon momentum buffers not in state dict
        self.muon_opt.reset()
        self.model.split_embed = False

        self.ws_short, self.ws_long = get_ws(0)
        self.batch_size = get_bs(0)
        self.model.yarn.reset()

    def get_state(self):
        return [copy.deepcopy(opt.state_dict()) for opt in self.optimizers]

# -----------------------------------------------------------------------------
# int main

@dataclass
class Hyperparameters:
    # data
    train_files: str = "data/fineweb10B/fineweb_train_*.bin" # input .bin to train on
    val_files: str = "data/fineweb10B/fineweb_val_*.bin" # input .bin to eval validation loss on
    val_tokens: int = 10485760 # how many tokens of validation data? it's important to keep this fixed for consistent comparisons
    # batch sizes
    train_bs_schedule: tuple = (131072, 262144, 393216, 524288, 
                                524288, 524288, 524288, 524288,
                                524288, 524288, 524288, 524288
                               )
    train_bs_extension: int = 32 * 2048 * 8
    train_max_seq_len: int = 128 * 16 * 2 # doubled to enable longer window sizes
    val_batch_size: int = 4 * 64 * 1024 * 8
    # optimization
    num_scheduled_iterations: int = 4700  # number of steps to complete lr and ws schedule
    num_extension_iterations: int = 40  # number of steps to continue training at final lr and ws
    num_iterations: int = num_scheduled_iterations + num_extension_iterations
    cooldown_frac: float = 0.70  # fraction of num_scheduled_iterations spent cooling down the learning rate
    split_embed_frac: float = 2/3/4
    # evaluation and logging
    run_id: str = f"{uuid.uuid4()}"
    val_loss_every: int = 250  # every how many steps to evaluate val loss? 0 for only at the end
    save_checkpoint: bool = False
    # attention masking
    block_size: int = 128
    ws_schedule: tuple = (3, 7, 11, 13,
                          15, 17, 19, 21,
                          23, 23, 23, 23)
    ws_final: int = 23 # set final validation ws, used for YaRN extension and short window size
    ws_validate_post_yarn_ext: int = 27 # extend long windows out even further after applying YaRN

args = Hyperparameters()

data_path = os.environ.get("DATA_PATH", ".")
args.train_files = os.path.join(data_path, args.train_files)
args.val_files = os.path.join(data_path, args.val_files)

# torchrun sets these env variables
rank = int(os.environ["RANK"])
world_size = int(os.environ["WORLD_SIZE"])
assert 8 % world_size == 0, "world_size must be a divisor of 8"
grad_accum_steps = 8 // world_size
assert torch.cuda.is_available()
device = torch.device("cuda", int(os.environ["LOCAL_RANK"]))
torch.cuda.set_device(device)
dist.init_process_group(backend="nccl", device_id=device)
dist.barrier()
master_process = (rank == 0) # this process will do logging, checkpointing etc.

# begin logging
logfile = None
if master_process:
    run_id = args.run_id
    os.makedirs("logs", exist_ok=True)
    logfile = f"logs/{run_id}.txt"
    print(logfile)
def print0(s, console=False):
    if master_process:
        with open(logfile, "a") as f:
            if console:
                print(s)
            print(s, file=f)

# begin by printing this file (the Python code)
print0(code)
print0("="*100)
# log information about the hardware/software environment this is running on
print0(f"Running Python {sys.version}")
print0(f"Running PyTorch {torch.version.__version__} compiled for CUDA {torch.version.cuda}")
print0(f"Running Triton version {triton.__version__}")

def nvidia_smi():
    import subprocess  # avoid top level import
    return subprocess.run(["nvidia-smi"], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True).stdout
print0(nvidia_smi())
print0("="*100)

model: nn.Module = GPT(
    vocab_size=50257,
    num_layers=16,
    num_heads=8,
    head_dim=128,
    model_dim=1024,
    max_seq_len=args.val_batch_size // (grad_accum_steps * world_size)
).cuda()
for m in model.modules():
    if isinstance(m, (nn.Embedding, nn.Linear)):
        m.weight.data = m.weight.data.bfloat16()
for param in model.parameters():
    dist.broadcast(param.detach(), 0)

model: nn.Module = torch.compile(model, dynamic=False, fullgraph=True)
training_manager = TrainingManager(model)

########################################
#            Warmup kernels            #
########################################
print0("Compiling model and warming up kernels (~7 minutes on first execution)", console=True)
# Warmup the training kernels, then re-initialize the state so we aren't cheating
initial_state = dict(model=copy.deepcopy(model.state_dict()),
                     optimizers=training_manager.get_state()) # save the initial state
train_loader = distributed_data_generator(args.train_files, args.train_bs_schedule[0], args.train_max_seq_len, grad_accum_steps=grad_accum_steps)
val_loader = distributed_data_generator(args.val_files, args.val_batch_size, -1, grad_accum_steps=grad_accum_steps, align_to_bos=False)

transition_steps = training_manager.get_transition_steps()
warmup_steps = sorted(set(s + offset for s in transition_steps for offset in [-1, 0, 1] if s + offset >= 0))
print0(f"Sampling steps {warmup_steps} for warmup", console=True)
for step in warmup_steps:
    training_manager.advance_schedule(step)
    model.eval()
    with torch.no_grad():
        inputs, targets, cum_seqlens = next(val_loader)
        model(inputs, targets, cum_seqlens, training_manager.get_forward_args())
    model.train()
    for idx in range(grad_accum_steps):
        # enable gradient sync for the DistAdam optimizers on the last iteration before we step them
        if idx == grad_accum_steps - 1:
            training_manager.activate_hooks(step)
        send_args = training_manager.train_loader_send_args
        inputs, targets, cum_seqlens = train_loader.send(send_args)
        (model(inputs, targets, cum_seqlens, training_manager.get_forward_args()) / grad_accum_steps).backward()
    training_manager.step_optimizers(step)
print0("Resetting Model", console=True)
model.zero_grad(set_to_none=True)
model.load_state_dict(initial_state["model"])
training_manager.reset(initial_state["optimizers"])
del val_loader, train_loader, initial_state
model.train()

########################################
#        Training and validation       #
########################################
train_loader = distributed_data_generator(args.train_files, args.train_bs_schedule[0], args.train_max_seq_len, grad_accum_steps=grad_accum_steps)

gc.collect()

training_time_ms = 0
# start the clock
torch.cuda.synchronize()
t0 = time.perf_counter()
# begin training
train_steps = args.num_iterations
for step in range(train_steps + 1):
    last_step = (step == train_steps)
    training_manager.advance_schedule(step)
    # --------------- VALIDATION SECTION -----------------
    if last_step or (args.val_loss_every > 0 and step % args.val_loss_every == 0):
        if last_step:
            training_manager.apply_final_ws_ext()
        # stop the clock
        torch.cuda.synchronize()
        training_time_ms += 1000 * (time.perf_counter() - t0)
        model.eval()
        assert args.val_tokens % args.val_batch_size == 0
        val_steps = grad_accum_steps * args.val_tokens // args.val_batch_size
        val_loader = distributed_data_generator(args.val_files, args.val_batch_size, -1, grad_accum_steps=grad_accum_steps, align_to_bos=False)
        val_loss = 0
        with torch.no_grad():
            for _ in range(val_steps):
                inputs, targets, cum_seqlens = next(val_loader)
                val_loss += model(inputs, targets, cum_seqlens, training_manager.get_forward_args())
        val_loss /= val_steps
        del val_loader
        dist.reduce(val_loss, 0, op=dist.ReduceOp.AVG)
        print0(f"step:{step}/{train_steps} val_loss:{val_loss:.4f} train_time:{training_time_ms:.0f}ms step_avg:{training_time_ms/max(step, 1):.2f}ms", console=True)
        model.train()
        # start the clock again
        torch.cuda.synchronize()
        t0 = time.perf_counter()

    if last_step:
        if master_process and args.save_checkpoint:
            log = dict(step=step, code=code, model=model.state_dict(), optimizers=[opt.state_dict() for opt in optimizers])
            os.makedirs(f"logs/{run_id}", exist_ok=True)
            torch.save(log, f"logs/{run_id}/state_step{step:06d}.pt")
        # the last step only has the validation loop, so break to avoid training
        break

    # --------------- TRAINING SECTION -----------------
    for idx in range(grad_accum_steps):
        # enable gradient sync for the DistAdam optimizers on the last iteration before we step them
        if idx == grad_accum_steps - 1:
            training_manager.activate_hooks(step)
        send_args = training_manager.train_loader_send_args
        inputs, targets, cum_seqlens = train_loader.send(send_args)
        (model(inputs, targets, cum_seqlens, training_manager.get_forward_args()) / grad_accum_steps).backward()
    training_manager.step_optimizers(step)

    # logging
    approx_training_time_ms = training_time_ms + 1000 * (time.perf_counter() - t0)
    print0(f"step:{step+1}/{train_steps} train_time:{approx_training_time_ms:.0f}ms step_avg:{approx_training_time_ms/(step + 1):.2f}ms", console=True)

print0(f"peak memory allocated: {torch.cuda.max_memory_allocated() // 1024 // 1024} MiB "
       f"reserved: {torch.cuda.max_memory_reserved() // 1024 // 1024} MiB", console=True)
dist.destroy_process_group()

====================================================================================================
Running Python 3.10.12 (main, Feb  4 2025, 14:57:36) [GCC 11.4.0]
Running PyTorch 2.10.0.dev20251210+cu126 compiled for CUDA 12.6
Running Triton version 3.6.0
Thu Jan  1 03:36:56 2026       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.127.08             Driver Version: 550.127.08     CUDA Version: 12.6     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA H100 80GB HBM3          On  |   00000000:19:00.0 Off |                    0 |
| N/A   36C    P0            124W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA H100 80GB HBM3          On  |   00000000:3B:00.0 Off |                    0 |
| N/A   31C    P0            125W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   2  NVIDIA H100 80GB HBM3          On  |   00000000:4C:00.0 Off |                    0 |
| N/A   30C    P0            120W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   3  NVIDIA H100 80GB HBM3          On  |   00000000:5D:00.0 Off |                    0 |
| N/A   34C    P0            122W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   4  NVIDIA H100 80GB HBM3          On  |   00000000:9B:00.0 Off |                    0 |
| N/A   35C    P0            120W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   5  NVIDIA H100 80GB HBM3          On  |   00000000:BB:00.0 Off |                    0 |
| N/A   30C    P0            118W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   6  NVIDIA H100 80GB HBM3          On  |   00000000:CB:00.0 Off |                    0 |
| N/A   34C    P0            122W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   7  NVIDIA H100 80GB HBM3          On  |   00000000:DB:00.0 Off |                    0 |
| N/A   29C    P0            113W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+

====================================================================================================
74
Compiling model and warming up kernels (~7 minutes on first execution)
Sampling steps [0, 1, 391, 392, 393, 783, 784, 785, 1174, 1175, 1176, 1566, 1567, 1568, 1958, 1959, 1960, 2349, 2350, 2351, 2741, 2742, 2743, 3133, 3134, 3135] for warmup
Resetting Model
step:0/4740 val_loss:10.8318 train_time:1ms step_avg:0.95ms
step:1/4740 train_time:90ms step_avg:89.94ms
step:2/4740 train_time:161ms step_avg:80.52ms
step:3/4740 train_time:218ms step_avg:72.75ms
step:4/4740 train_time:303ms step_avg:75.76ms
step:5/4740 train_time:361ms step_avg:72.19ms
step:6/4740 train_time:446ms step_avg:74.25ms
step:7/4740 train_time:504ms step_avg:72.02ms
step:8/4740 train_time:589ms step_avg:73.64ms
step:9/4740 train_time:647ms step_avg:71.86ms
step:10/4740 train_time:731ms step_avg:73.14ms
step:11/4740 train_time:790ms step_avg:71.79ms
step:12/4740 train_time:875ms step_avg:72.93ms
step:13/4740 train_time:933ms step_avg:71.80ms
step:14/4740 train_time:1021ms step_avg:72.90ms
step:15/4740 train_time:1080ms step_avg:71.98ms
step:16/4740 train_time:1167ms step_avg:72.97ms
step:17/4740 train_time:1226ms step_avg:72.14ms
step:18/4740 train_time:1312ms step_avg:72.88ms
step:19/4740 train_time:1371ms step_avg:72.14ms
step:20/4740 train_time:1457ms step_avg:72.86ms
step:21/4740 train_time:1517ms step_avg:72.22ms
step:22/4740 train_time:1602ms step_avg:72.83ms
step:23/4740 train_time:1661ms step_avg:72.23ms
step:24/4740 train_time:1746ms step_avg:72.77ms
step:25/4740 train_time:1805ms step_avg:72.19ms
step:26/4740 train_time:1890ms step_avg:72.71ms
step:27/4740 train_time:1949ms step_avg:72.18ms
step:28/4740 train_time:2035ms step_avg:72.69ms
step:29/4740 train_time:2095ms step_avg:72.25ms
step:30/4740 train_time:2182ms step_avg:72.72ms
step:31/4740 train_time:2242ms step_avg:72.32ms
step:32/4740 train_time:2327ms step_avg:72.72ms
step:33/4740 train_time:2386ms step_avg:72.30ms
step:34/4740 train_time:2471ms step_avg:72.68ms
step:35/4740 train_time:2530ms step_avg:72.27ms
step:36/4740 train_time:2616ms step_avg:72.66ms
step:37/4740 train_time:2674ms step_avg:72.28ms
step:38/4740 train_time:2760ms step_avg:72.63ms
step:39/4740 train_time:2819ms step_avg:72.27ms
step:40/4740 train_time:2904ms step_avg:72.61ms
step:41/4740 train_time:2964ms step_avg:72.29ms
step:42/4740 train_time:3050ms step_avg:72.61ms
step:43/4740 train_time:3109ms step_avg:72.31ms
step:44/4740 train_time:3197ms step_avg:72.66ms
step:45/4740 train_time:3255ms step_avg:72.33ms
step:46/4740 train_time:3340ms step_avg:72.62ms
step:47/4740 train_time:3400ms step_avg:72.34ms
step:48/4740 train_time:3486ms step_avg:72.62ms
step:49/4740 train_time:3546ms step_avg:72.36ms
step:50/4740 train_time:3632ms step_avg:72.64ms
step:51/4740 train_time:3691ms step_avg:72.36ms
step:52/4740 train_time:3776ms step_avg:72.62ms
step:53/4740 train_time:3835ms step_avg:72.36ms
step:54/4740 train_time:3921ms step_avg:72.60ms
step:55/4740 train_time:3979ms step_avg:72.34ms
step:56/4740 train_time:4064ms step_avg:72.58ms
step:57/4740 train_time:4124ms step_avg:72.34ms
step:58/4740 train_time:4210ms step_avg:72.59ms
step:59/4740 train_time:4269ms step_avg:72.35ms
step:60/4740 train_time:4354ms step_avg:72.57ms
step:61/4740 train_time:4413ms step_avg:72.34ms
step:62/4740 train_time:4498ms step_avg:72.55ms
step:63/4740 train_time:4558ms step_avg:72.34ms
step:64/4740 train_time:4643ms step_avg:72.55ms
step:65/4740 train_time:4702ms step_avg:72.35ms
step:66/4740 train_time:4788ms step_avg:72.55ms
step:67/4740 train_time:4847ms step_avg:72.34ms
step:68/4740 train_time:4933ms step_avg:72.54ms
step:69/4740 train_time:4991ms step_avg:72.33ms
step:70/4740 train_time:5077ms step_avg:72.53ms
step:71/4740 train_time:5136ms step_avg:72.34ms
step:72/4740 train_time:5223ms step_avg:72.53ms
step:73/4740 train_time:5283ms step_avg:72.37ms
step:74/4740 train_time:5369ms step_avg:72.55ms
step:75/4740 train_time:5428ms step_avg:72.37ms
step:76/4740 train_time:5514ms step_avg:72.55ms
step:77/4740 train_time:5572ms step_avg:72.36ms
step:78/4740 train_time:5659ms step_avg:72.55ms
step:79/4740 train_time:5717ms step_avg:72.37ms
step:80/4740 train_time:5803ms step_avg:72.54ms
step:81/4740 train_time:5861ms step_avg:72.36ms
step:82/4740 train_time:5948ms step_avg:72.53ms
step:83/4740 train_time:6006ms step_avg:72.37ms
step:84/4740 train_time:6093ms step_avg:72.53ms
step:85/4740 train_time:6151ms step_avg:72.36ms
step:86/4740 train_time:6239ms step_avg:72.54ms
step:87/4740 train_time:6298ms step_avg:72.39ms
step:88/4740 train_time:6384ms step_avg:72.54ms
step:89/4740 train_time:6443ms step_avg:72.39ms
step:90/4740 train_time:6528ms step_avg:72.54ms
step:91/4740 train_time:6587ms step_avg:72.38ms
step:92/4740 train_time:6673ms step_avg:72.53ms
step:93/4740 train_time:6732ms step_avg:72.38ms
step:94/4740 train_time:6817ms step_avg:72.53ms
step:95/4740 train_time:6876ms step_avg:72.38ms
step:96/4740 train_time:6962ms step_avg:72.52ms
step:97/4740 train_time:7021ms step_avg:72.38ms
step:98/4740 train_time:7106ms step_avg:72.51ms
step:99/4740 train_time:7165ms step_avg:72.38ms
step:100/4740 train_time:7251ms step_avg:72.51ms
step:101/4740 train_time:7309ms step_avg:72.37ms
step:102/4740 train_time:7395ms step_avg:72.50ms
step:103/4740 train_time:7454ms step_avg:72.37ms
step:104/4740 train_time:7540ms step_avg:72.50ms
step:105/4740 train_time:7599ms step_avg:72.37ms
step:106/4740 train_time:7685ms step_avg:72.50ms
step:107/4740 train_time:7744ms step_avg:72.38ms
step:108/4740 train_time:7831ms step_avg:72.51ms
step:109/4740 train_time:7890ms step_avg:72.38ms
step:110/4740 train_time:7976ms step_avg:72.51ms
step:111/4740 train_time:8034ms step_avg:72.38ms
step:112/4740 train_time:8121ms step_avg:72.51ms
step:113/4740 train_time:8179ms step_avg:72.38ms
step:114/4740 train_time:8266ms step_avg:72.51ms
step:115/4740 train_time:8325ms step_avg:72.39ms
step:116/4740 train_time:8411ms step_avg:72.51ms
step:117/4740 train_time:8469ms step_avg:72.39ms
step:118/4740 train_time:8555ms step_avg:72.50ms
step:119/4740 train_time:8614ms step_avg:72.38ms
step:120/4740 train_time:8699ms step_avg:72.49ms
step:121/4740 train_time:8759ms step_avg:72.39ms
step:122/4740 train_time:8845ms step_avg:72.50ms
step:123/4740 train_time:8904ms step_avg:72.39ms
step:124/4740 train_time:8990ms step_avg:72.50ms
step:125/4740 train_time:9048ms step_avg:72.39ms
step:126/4740 train_time:9134ms step_avg:72.49ms
step:127/4740 train_time:9193ms step_avg:72.39ms
step:128/4740 train_time:9280ms step_avg:72.50ms
step:129/4740 train_time:9339ms step_avg:72.39ms
step:130/4740 train_time:9425ms step_avg:72.50ms
step:131/4740 train_time:9483ms step_avg:72.39ms
step:132/4740 train_time:9569ms step_avg:72.50ms
step:133/4740 train_time:9628ms step_avg:72.39ms
step:134/4740 train_time:9714ms step_avg:72.49ms
step:135/4740 train_time:9772ms step_avg:72.39ms
step:136/4740 train_time:9858ms step_avg:72.48ms
step:137/4740 train_time:9917ms step_avg:72.38ms
step:138/4740 train_time:10002ms step_avg:72.48ms
step:139/4740 train_time:10061ms step_avg:72.38ms
step:140/4740 train_time:10147ms step_avg:72.48ms
step:141/4740 train_time:10206ms step_avg:72.39ms
step:142/4740 train_time:10292ms step_avg:72.48ms
step:143/4740 train_time:10350ms step_avg:72.38ms
step:144/4740 train_time:10436ms step_avg:72.47ms
step:145/4740 train_time:10495ms step_avg:72.38ms
step:146/4740 train_time:10580ms step_avg:72.47ms
step:147/4740 train_time:10639ms step_avg:72.37ms
step:148/4740 train_time:10725ms step_avg:72.47ms
step:149/4740 train_time:10783ms step_avg:72.37ms
step:150/4740 train_time:10869ms step_avg:72.46ms
step:151/4740 train_time:10927ms step_avg:72.37ms
step:152/4740 train_time:11013ms step_avg:72.45ms
step:153/4740 train_time:11070ms step_avg:72.36ms
step:154/4740 train_time:11157ms step_avg:72.45ms
step:155/4740 train_time:11217ms step_avg:72.37ms
step:156/4740 train_time:11302ms step_avg:72.45ms
step:157/4740 train_time:11361ms step_avg:72.36ms
step:158/4740 train_time:11447ms step_avg:72.45ms
step:159/4740 train_time:11505ms step_avg:72.36ms
step:160/4740 train_time:11591ms step_avg:72.44ms
step:161/4740 train_time:11649ms step_avg:72.36ms
step:162/4740 train_time:11736ms step_avg:72.44ms
step:163/4740 train_time:11794ms step_avg:72.36ms
step:164/4740 train_time:11880ms step_avg:72.44ms
step:165/4740 train_time:11939ms step_avg:72.36ms
step:166/4740 train_time:12024ms step_avg:72.44ms
step:167/4740 train_time:12084ms step_avg:72.36ms
step:168/4740 train_time:12169ms step_avg:72.43ms
step:169/4740 train_time:12227ms step_avg:72.35ms
step:170/4740 train_time:12313ms step_avg:72.43ms
step:171/4740 train_time:12371ms step_avg:72.34ms
step:172/4740 train_time:12456ms step_avg:72.42ms
step:173/4740 train_time:12515ms step_avg:72.34ms
step:174/4740 train_time:12601ms step_avg:72.42ms
step:175/4740 train_time:12660ms step_avg:72.35ms
step:176/4740 train_time:12747ms step_avg:72.42ms
step:177/4740 train_time:12805ms step_avg:72.35ms
step:178/4740 train_time:12891ms step_avg:72.42ms
step:179/4740 train_time:12949ms step_avg:72.34ms
step:180/4740 train_time:13035ms step_avg:72.42ms
step:181/4740 train_time:13093ms step_avg:72.34ms
step:182/4740 train_time:13179ms step_avg:72.41ms
step:183/4740 train_time:13237ms step_avg:72.33ms
step:184/4740 train_time:13323ms step_avg:72.41ms
step:185/4740 train_time:13382ms step_avg:72.33ms
step:186/4740 train_time:13466ms step_avg:72.40ms
step:187/4740 train_time:13525ms step_avg:72.33ms
step:188/4740 train_time:13611ms step_avg:72.40ms
step:189/4740 train_time:13669ms step_avg:72.32ms
step:190/4740 train_time:13755ms step_avg:72.39ms
step:191/4740 train_time:13814ms step_avg:72.32ms
step:192/4740 train_time:13899ms step_avg:72.39ms
step:193/4740 train_time:13958ms step_avg:72.32ms
step:194/4740 train_time:14043ms step_avg:72.38ms
step:195/4740 train_time:14101ms step_avg:72.31ms
step:196/4740 train_time:14187ms step_avg:72.38ms
step:197/4740 train_time:14245ms step_avg:72.31ms
step:198/4740 train_time:14331ms step_avg:72.38ms
step:199/4740 train_time:14390ms step_avg:72.31ms
step:200/4740 train_time:14475ms step_avg:72.37ms
step:201/4740 train_time:14534ms step_avg:72.31ms
step:202/4740 train_time:14619ms step_avg:72.37ms
step:203/4740 train_time:14677ms step_avg:72.30ms
step:204/4740 train_time:14763ms step_avg:72.37ms
step:205/4740 train_time:14822ms step_avg:72.30ms
step:206/4740 train_time:14907ms step_avg:72.37ms
step:207/4740 train_time:14966ms step_avg:72.30ms
step:208/4740 train_time:15053ms step_avg:72.37ms
step:209/4740 train_time:15111ms step_avg:72.30ms
step:210/4740 train_time:15197ms step_avg:72.37ms
step:211/4740 train_time:15255ms step_avg:72.30ms
step:212/4740 train_time:15340ms step_avg:72.36ms
step:213/4740 train_time:15400ms step_avg:72.30ms
step:214/4740 train_time:15485ms step_avg:72.36ms
step:215/4740 train_time:15543ms step_avg:72.29ms
step:216/4740 train_time:15627ms step_avg:72.35ms
step:217/4740 train_time:15686ms step_avg:72.28ms
step:218/4740 train_time:15771ms step_avg:72.34ms
step:219/4740 train_time:15829ms step_avg:72.28ms
step:220/4740 train_time:15915ms step_avg:72.34ms
step:221/4740 train_time:15974ms step_avg:72.28ms
step:222/4740 train_time:16059ms step_avg:72.34ms
step:223/4740 train_time:16118ms step_avg:72.28ms
step:224/4740 train_time:16203ms step_avg:72.34ms
step:225/4740 train_time:16262ms step_avg:72.27ms
step:226/4740 train_time:16347ms step_avg:72.33ms
step:227/4740 train_time:16405ms step_avg:72.27ms
step:228/4740 train_time:16490ms step_avg:72.33ms
step:229/4740 train_time:16549ms step_avg:72.26ms
step:230/4740 train_time:16634ms step_avg:72.32ms
step:231/4740 train_time:16692ms step_avg:72.26ms
step:232/4740 train_time:16777ms step_avg:72.31ms
step:233/4740 train_time:16835ms step_avg:72.25ms
step:234/4740 train_time:16921ms step_avg:72.31ms
step:235/4740 train_time:16979ms step_avg:72.25ms
step:236/4740 train_time:17065ms step_avg:72.31ms
step:237/4740 train_time:17123ms step_avg:72.25ms
step:238/4740 train_time:17209ms step_avg:72.31ms
step:239/4740 train_time:17268ms step_avg:72.25ms
step:240/4740 train_time:17353ms step_avg:72.30ms
step:241/4740 train_time:17411ms step_avg:72.25ms
step:242/4740 train_time:17497ms step_avg:72.30ms
step:243/4740 train_time:17555ms step_avg:72.24ms
step:244/4740 train_time:17641ms step_avg:72.30ms
step:245/4740 train_time:17699ms step_avg:72.24ms
step:246/4740 train_time:17785ms step_avg:72.30ms
step:247/4740 train_time:17844ms step_avg:72.24ms
step:248/4740 train_time:17930ms step_avg:72.30ms
step:249/4740 train_time:17988ms step_avg:72.24ms
step:250/4740 train_time:18072ms step_avg:72.29ms
step:250/4740 val_loss:4.5266 train_time:18117ms step_avg:72.47ms
step:251/4740 train_time:18142ms step_avg:72.28ms
step:252/4740 train_time:18217ms step_avg:72.29ms
step:253/4740 train_time:18275ms step_avg:72.23ms
step:254/4740 train_time:18360ms step_avg:72.28ms
step:255/4740 train_time:18421ms step_avg:72.24ms
step:256/4740 train_time:18508ms step_avg:72.30ms
step:257/4740 train_time:18568ms step_avg:72.25ms
step:258/4740 train_time:18655ms step_avg:72.31ms
step:259/4740 train_time:18713ms step_avg:72.25ms
step:260/4740 train_time:18798ms step_avg:72.30ms
step:261/4740 train_time:18856ms step_avg:72.25ms
step:262/4740 train_time:18941ms step_avg:72.29ms
step:263/4740 train_time:18999ms step_avg:72.24ms
step:264/4740 train_time:19084ms step_avg:72.29ms
step:265/4740 train_time:19141ms step_avg:72.23ms
step:266/4740 train_time:19226ms step_avg:72.28ms
step:267/4740 train_time:19284ms step_avg:72.22ms
step:268/4740 train_time:19369ms step_avg:72.27ms
step:269/4740 train_time:19429ms step_avg:72.23ms
step:270/4740 train_time:19516ms step_avg:72.28ms
step:271/4740 train_time:19576ms step_avg:72.24ms
step:272/4740 train_time:19661ms step_avg:72.28ms
step:273/4740 train_time:19720ms step_avg:72.23ms
step:274/4740 train_time:19806ms step_avg:72.28ms
step:275/4740 train_time:19864ms step_avg:72.23ms
step:276/4740 train_time:19948ms step_avg:72.28ms
step:277/4740 train_time:20006ms step_avg:72.22ms
step:278/4740 train_time:20093ms step_avg:72.28ms
step:279/4740 train_time:20151ms step_avg:72.23ms
step:280/4740 train_time:20236ms step_avg:72.27ms
step:281/4740 train_time:20294ms step_avg:72.22ms
step:282/4740 train_time:20380ms step_avg:72.27ms
step:283/4740 train_time:20439ms step_avg:72.22ms
step:284/4740 train_time:20525ms step_avg:72.27ms
step:285/4740 train_time:20584ms step_avg:72.23ms
step:286/4740 train_time:20670ms step_avg:72.27ms
step:287/4740 train_time:20728ms step_avg:72.22ms
step:288/4740 train_time:20814ms step_avg:72.27ms
step:289/4740 train_time:20872ms step_avg:72.22ms
step:290/4740 train_time:20957ms step_avg:72.27ms
step:291/4740 train_time:21016ms step_avg:72.22ms
step:292/4740 train_time:21102ms step_avg:72.27ms
step:293/4740 train_time:21159ms step_avg:72.22ms
step:294/4740 train_time:21245ms step_avg:72.26ms
step:295/4740 train_time:21303ms step_avg:72.21ms
step:296/4740 train_time:21387ms step_avg:72.25ms
step:297/4740 train_time:21445ms step_avg:72.21ms
step:298/4740 train_time:21531ms step_avg:72.25ms
step:299/4740 train_time:21590ms step_avg:72.21ms
step:300/4740 train_time:21675ms step_avg:72.25ms
step:301/4740 train_time:21734ms step_avg:72.21ms
step:302/4740 train_time:21819ms step_avg:72.25ms
step:303/4740 train_time:21878ms step_avg:72.20ms
step:304/4740 train_time:21963ms step_avg:72.25ms
step:305/4740 train_time:22021ms step_avg:72.20ms
step:306/4740 train_time:22106ms step_avg:72.24ms
step:307/4740 train_time:22164ms step_avg:72.20ms
step:308/4740 train_time:22250ms step_avg:72.24ms
step:309/4740 train_time:22307ms step_avg:72.19ms
step:310/4740 train_time:22393ms step_avg:72.23ms
step:311/4740 train_time:22451ms step_avg:72.19ms
step:312/4740 train_time:22536ms step_avg:72.23ms
step:313/4740 train_time:22595ms step_avg:72.19ms
step:314/4740 train_time:22680ms step_avg:72.23ms
step:315/4740 train_time:22739ms step_avg:72.19ms
step:316/4740 train_time:22824ms step_avg:72.23ms
step:317/4740 train_time:22883ms step_avg:72.19ms
step:318/4740 train_time:22968ms step_avg:72.23ms
step:319/4740 train_time:23026ms step_avg:72.18ms
step:320/4740 train_time:23112ms step_avg:72.22ms
step:321/4740 train_time:23170ms step_avg:72.18ms
step:322/4740 train_time:23255ms step_avg:72.22ms
step:323/4740 train_time:23313ms step_avg:72.18ms
step:324/4740 train_time:23398ms step_avg:72.21ms
step:325/4740 train_time:23455ms step_avg:72.17ms
step:326/4740 train_time:23540ms step_avg:72.21ms
step:327/4740 train_time:23599ms step_avg:72.17ms
step:328/4740 train_time:23684ms step_avg:72.21ms
step:329/4740 train_time:23744ms step_avg:72.17ms
step:330/4740 train_time:23829ms step_avg:72.21ms
step:331/4740 train_time:23886ms step_avg:72.16ms
step:332/4740 train_time:23972ms step_avg:72.21ms
step:333/4740 train_time:24030ms step_avg:72.16ms
step:334/4740 train_time:24116ms step_avg:72.20ms
step:335/4740 train_time:24174ms step_avg:72.16ms
step:336/4740 train_time:24258ms step_avg:72.20ms
step:337/4740 train_time:24317ms step_avg:72.16ms
step:338/4740 train_time:24402ms step_avg:72.19ms
step:339/4740 train_time:24461ms step_avg:72.16ms
step:340/4740 train_time:24546ms step_avg:72.19ms
step:341/4740 train_time:24604ms step_avg:72.15ms
step:342/4740 train_time:24689ms step_avg:72.19ms
step:343/4740 train_time:24747ms step_avg:72.15ms
step:344/4740 train_time:24832ms step_avg:72.18ms
step:345/4740 train_time:24890ms step_avg:72.15ms
step:346/4740 train_time:24976ms step_avg:72.18ms
step:347/4740 train_time:25034ms step_avg:72.14ms
step:348/4740 train_time:25119ms step_avg:72.18ms
step:349/4740 train_time:25177ms step_avg:72.14ms
step:350/4740 train_time:25261ms step_avg:72.18ms
step:351/4740 train_time:25320ms step_avg:72.14ms
step:352/4740 train_time:25405ms step_avg:72.17ms
step:353/4740 train_time:25463ms step_avg:72.13ms
step:354/4740 train_time:25549ms step_avg:72.17ms
step:355/4740 train_time:25606ms step_avg:72.13ms
step:356/4740 train_time:25692ms step_avg:72.17ms
step:357/4740 train_time:25751ms step_avg:72.13ms
step:358/4740 train_time:25837ms step_avg:72.17ms
step:359/4740 train_time:25895ms step_avg:72.13ms
step:360/4740 train_time:25981ms step_avg:72.17ms
step:361/4740 train_time:26040ms step_avg:72.13ms
step:362/4740 train_time:26125ms step_avg:72.17ms
step:363/4740 train_time:26183ms step_avg:72.13ms
step:364/4740 train_time:26268ms step_avg:72.16ms
step:365/4740 train_time:26326ms step_avg:72.13ms
step:366/4740 train_time:26412ms step_avg:72.16ms
step:367/4740 train_time:26471ms step_avg:72.13ms
step:368/4740 train_time:26556ms step_avg:72.16ms
step:369/4740 train_time:26614ms step_avg:72.12ms
step:370/4740 train_time:26699ms step_avg:72.16ms
step:371/4740 train_time:26758ms step_avg:72.12ms
step:372/4740 train_time:26845ms step_avg:72.16ms
step:373/4740 train_time:26903ms step_avg:72.13ms
step:374/4740 train_time:26988ms step_avg:72.16ms
step:375/4740 train_time:27047ms step_avg:72.12ms
step:376/4740 train_time:27133ms step_avg:72.16ms
step:377/4740 train_time:27191ms step_avg:72.12ms
step:378/4740 train_time:27276ms step_avg:72.16ms
step:379/4740 train_time:27334ms step_avg:72.12ms
step:380/4740 train_time:27419ms step_avg:72.15ms
step:381/4740 train_time:27477ms step_avg:72.12ms
step:382/4740 train_time:27562ms step_avg:72.15ms
step:383/4740 train_time:27621ms step_avg:72.12ms
step:384/4740 train_time:27707ms step_avg:72.15ms
step:385/4740 train_time:27765ms step_avg:72.12ms
step:386/4740 train_time:27851ms step_avg:72.15ms
step:387/4740 train_time:27909ms step_avg:72.12ms
step:388/4740 train_time:27994ms step_avg:72.15ms
step:389/4740 train_time:28053ms step_avg:72.11ms
step:390/4740 train_time:28138ms step_avg:72.15ms
step:391/4740 train_time:28196ms step_avg:72.11ms
step:392/4740 train_time:28282ms step_avg:72.15ms
step:393/4740 train_time:28354ms step_avg:72.15ms
step:394/4740 train_time:28511ms step_avg:72.36ms
step:395/4740 train_time:28610ms step_avg:72.43ms
step:396/4740 train_time:28769ms step_avg:72.65ms
step:397/4740 train_time:28869ms step_avg:72.72ms
step:398/4740 train_time:29027ms step_avg:72.93ms
step:399/4740 train_time:29127ms step_avg:73.00ms
step:400/4740 train_time:29285ms step_avg:73.21ms
step:401/4740 train_time:29385ms step_avg:73.28ms
step:402/4740 train_time:29543ms step_avg:73.49ms
step:403/4740 train_time:29641ms step_avg:73.55ms
step:404/4740 train_time:29800ms step_avg:73.76ms
step:405/4740 train_time:29899ms step_avg:73.82ms
step:406/4740 train_time:30057ms step_avg:74.03ms
step:407/4740 train_time:30157ms step_avg:74.10ms
step:408/4740 train_time:30316ms step_avg:74.30ms
step:409/4740 train_time:30415ms step_avg:74.37ms
step:410/4740 train_time:30573ms step_avg:74.57ms
step:411/4740 train_time:30672ms step_avg:74.63ms
step:412/4740 train_time:30831ms step_avg:74.83ms
step:413/4740 train_time:30930ms step_avg:74.89ms
step:414/4740 train_time:31090ms step_avg:75.10ms
step:415/4740 train_time:31190ms step_avg:75.16ms
step:416/4740 train_time:31348ms step_avg:75.36ms
step:417/4740 train_time:31447ms step_avg:75.41ms
step:418/4740 train_time:31606ms step_avg:75.61ms
step:419/4740 train_time:31706ms step_avg:75.67ms
step:420/4740 train_time:31865ms step_avg:75.87ms
step:421/4740 train_time:31965ms step_avg:75.93ms
step:422/4740 train_time:32123ms step_avg:76.12ms
step:423/4740 train_time:32222ms step_avg:76.17ms
step:424/4740 train_time:32380ms step_avg:76.37ms
step:425/4740 train_time:32479ms step_avg:76.42ms
step:426/4740 train_time:32637ms step_avg:76.61ms
step:427/4740 train_time:32735ms step_avg:76.66ms
step:428/4740 train_time:32893ms step_avg:76.85ms
step:429/4740 train_time:32992ms step_avg:76.90ms
step:430/4740 train_time:33151ms step_avg:77.10ms
step:431/4740 train_time:33251ms step_avg:77.15ms
step:432/4740 train_time:33410ms step_avg:77.34ms
step:433/4740 train_time:33508ms step_avg:77.39ms
step:434/4740 train_time:33667ms step_avg:77.57ms
step:435/4740 train_time:33767ms step_avg:77.62ms
step:436/4740 train_time:33926ms step_avg:77.81ms
step:437/4740 train_time:34026ms step_avg:77.86ms
step:438/4740 train_time:34185ms step_avg:78.05ms
step:439/4740 train_time:34287ms step_avg:78.10ms
step:440/4740 train_time:34444ms step_avg:78.28ms
step:441/4740 train_time:34544ms step_avg:78.33ms
step:442/4740 train_time:34702ms step_avg:78.51ms
step:443/4740 train_time:34801ms step_avg:78.56ms
step:444/4740 train_time:34960ms step_avg:78.74ms
step:445/4740 train_time:35059ms step_avg:78.78ms
step:446/4740 train_time:35218ms step_avg:78.96ms
step:447/4740 train_time:35318ms step_avg:79.01ms
step:448/4740 train_time:35476ms step_avg:79.19ms
step:449/4740 train_time:35575ms step_avg:79.23ms
step:450/4740 train_time:35732ms step_avg:79.40ms
step:451/4740 train_time:35831ms step_avg:79.45ms
step:452/4740 train_time:35991ms step_avg:79.63ms
step:453/4740 train_time:36091ms step_avg:79.67ms
step:454/4740 train_time:36249ms step_avg:79.84ms
step:455/4740 train_time:36349ms step_avg:79.89ms
step:456/4740 train_time:36507ms step_avg:80.06ms
step:457/4740 train_time:36607ms step_avg:80.10ms
step:458/4740 train_time:36765ms step_avg:80.27ms
step:459/4740 train_time:36865ms step_avg:80.32ms
step:460/4740 train_time:37024ms step_avg:80.49ms
step:461/4740 train_time:37124ms step_avg:80.53ms
step:462/4740 train_time:37284ms step_avg:80.70ms
step:463/4740 train_time:37382ms step_avg:80.74ms
step:464/4740 train_time:37541ms step_avg:80.91ms
step:465/4740 train_time:37640ms step_avg:80.95ms
step:466/4740 train_time:37799ms step_avg:81.11ms
step:467/4740 train_time:37898ms step_avg:81.15ms
step:468/4740 train_time:38056ms step_avg:81.32ms
step:469/4740 train_time:38156ms step_avg:81.36ms
step:470/4740 train_time:38317ms step_avg:81.52ms
step:471/4740 train_time:38415ms step_avg:81.56ms
step:472/4740 train_time:38574ms step_avg:81.72ms
step:473/4740 train_time:38673ms step_avg:81.76ms
step:474/4740 train_time:38831ms step_avg:81.92ms
step:475/4740 train_time:38931ms step_avg:81.96ms
step:476/4740 train_time:39091ms step_avg:82.12ms
step:477/4740 train_time:39190ms step_avg:82.16ms
step:478/4740 train_time:39348ms step_avg:82.32ms
step:479/4740 train_time:39447ms step_avg:82.35ms
step:480/4740 train_time:39605ms step_avg:82.51ms
step:481/4740 train_time:39707ms step_avg:82.55ms
step:482/4740 train_time:39865ms step_avg:82.71ms
step:483/4740 train_time:39966ms step_avg:82.75ms
step:484/4740 train_time:40125ms step_avg:82.90ms
step:485/4740 train_time:40225ms step_avg:82.94ms
step:486/4740 train_time:40384ms step_avg:83.10ms
step:487/4740 train_time:40484ms step_avg:83.13ms
step:488/4740 train_time:40642ms step_avg:83.28ms
step:489/4740 train_time:40741ms step_avg:83.31ms
step:490/4740 train_time:40901ms step_avg:83.47ms
step:491/4740 train_time:41002ms step_avg:83.51ms
step:492/4740 train_time:41161ms step_avg:83.66ms
step:493/4740 train_time:41261ms step_avg:83.69ms
step:494/4740 train_time:41419ms step_avg:83.85ms
step:495/4740 train_time:41518ms step_avg:83.88ms
step:496/4740 train_time:41677ms step_avg:84.03ms
step:497/4740 train_time:41777ms step_avg:84.06ms
step:498/4740 train_time:41935ms step_avg:84.21ms
step:499/4740 train_time:42036ms step_avg:84.24ms
step:500/4740 train_time:42193ms step_avg:84.39ms
step:500/4740 val_loss:4.0326 train_time:42265ms step_avg:84.53ms
step:501/4740 train_time:42295ms step_avg:84.42ms
step:502/4740 train_time:42454ms step_avg:84.57ms
step:503/4740 train_time:42556ms step_avg:84.60ms
step:504/4740 train_time:42725ms step_avg:84.77ms
step:505/4740 train_time:42826ms step_avg:84.80ms
step:506/4740 train_time:42983ms step_avg:84.95ms
step:507/4740 train_time:43081ms step_avg:84.97ms
step:508/4740 train_time:43238ms step_avg:85.11ms
step:509/4740 train_time:43337ms step_avg:85.14ms
step:510/4740 train_time:43496ms step_avg:85.29ms
step:511/4740 train_time:43599ms step_avg:85.32ms
step:512/4740 train_time:43761ms step_avg:85.47ms
step:513/4740 train_time:43862ms step_avg:85.50ms
step:514/4740 train_time:44020ms step_avg:85.64ms
step:515/4740 train_time:44119ms step_avg:85.67ms
step:516/4740 train_time:44276ms step_avg:85.81ms
step:517/4740 train_time:44374ms step_avg:85.83ms
step:518/4740 train_time:44534ms step_avg:85.97ms
step:519/4740 train_time:44634ms step_avg:86.00ms
step:520/4740 train_time:44796ms step_avg:86.15ms
step:521/4740 train_time:44895ms step_avg:86.17ms
step:522/4740 train_time:45053ms step_avg:86.31ms
step:523/4740 train_time:45152ms step_avg:86.33ms
step:524/4740 train_time:45309ms step_avg:86.47ms
step:525/4740 train_time:45408ms step_avg:86.49ms
step:526/4740 train_time:45567ms step_avg:86.63ms
step:527/4740 train_time:45668ms step_avg:86.66ms
step:528/4740 train_time:45826ms step_avg:86.79ms
step:529/4740 train_time:45927ms step_avg:86.82ms
step:530/4740 train_time:46085ms step_avg:86.95ms
step:531/4740 train_time:46183ms step_avg:86.97ms
step:532/4740 train_time:46341ms step_avg:87.11ms
step:533/4740 train_time:46441ms step_avg:87.13ms
step:534/4740 train_time:46600ms step_avg:87.27ms
step:535/4740 train_time:46702ms step_avg:87.29ms
step:536/4740 train_time:46861ms step_avg:87.43ms
step:537/4740 train_time:46961ms step_avg:87.45ms
step:538/4740 train_time:47119ms step_avg:87.58ms
step:539/4740 train_time:47220ms step_avg:87.61ms
step:540/4740 train_time:47378ms step_avg:87.74ms
step:541/4740 train_time:47478ms step_avg:87.76ms
step:542/4740 train_time:47637ms step_avg:87.89ms
step:543/4740 train_time:47738ms step_avg:87.91ms
step:544/4740 train_time:47896ms step_avg:88.04ms
step:545/4740 train_time:47995ms step_avg:88.06ms
step:546/4740 train_time:48153ms step_avg:88.19ms
step:547/4740 train_time:48252ms step_avg:88.21ms
step:548/4740 train_time:48409ms step_avg:88.34ms
step:549/4740 train_time:48510ms step_avg:88.36ms
step:550/4740 train_time:48668ms step_avg:88.49ms
step:551/4740 train_time:48767ms step_avg:88.51ms
step:552/4740 train_time:48925ms step_avg:88.63ms
step:553/4740 train_time:49025ms step_avg:88.65ms
step:554/4740 train_time:49183ms step_avg:88.78ms
step:555/4740 train_time:49283ms step_avg:88.80ms
step:556/4740 train_time:49442ms step_avg:88.93ms
step:557/4740 train_time:49542ms step_avg:88.94ms
step:558/4740 train_time:49700ms step_avg:89.07ms
step:559/4740 train_time:49803ms step_avg:89.09ms
step:560/4740 train_time:49961ms step_avg:89.22ms
step:561/4740 train_time:50061ms step_avg:89.24ms
step:562/4740 train_time:50220ms step_avg:89.36ms
step:563/4740 train_time:50321ms step_avg:89.38ms
step:564/4740 train_time:50480ms step_avg:89.50ms
step:565/4740 train_time:50579ms step_avg:89.52ms
step:566/4740 train_time:50737ms step_avg:89.64ms
step:567/4740 train_time:50836ms step_avg:89.66ms
step:568/4740 train_time:50995ms step_avg:89.78ms
step:569/4740 train_time:51093ms step_avg:89.79ms
step:570/4740 train_time:51252ms step_avg:89.92ms
step:571/4740 train_time:51351ms step_avg:89.93ms
step:572/4740 train_time:51510ms step_avg:90.05ms
step:573/4740 train_time:51610ms step_avg:90.07ms
step:574/4740 train_time:51767ms step_avg:90.19ms
step:575/4740 train_time:51868ms step_avg:90.20ms
step:576/4740 train_time:52026ms step_avg:90.32ms
step:577/4740 train_time:52126ms step_avg:90.34ms
step:578/4740 train_time:52284ms step_avg:90.46ms
step:579/4740 train_time:52383ms step_avg:90.47ms
step:580/4740 train_time:52542ms step_avg:90.59ms
step:581/4740 train_time:52641ms step_avg:90.60ms
step:582/4740 train_time:52800ms step_avg:90.72ms
step:583/4740 train_time:52900ms step_avg:90.74ms
step:584/4740 train_time:53058ms step_avg:90.85ms
step:585/4740 train_time:53158ms step_avg:90.87ms
step:586/4740 train_time:53316ms step_avg:90.98ms
step:587/4740 train_time:53417ms step_avg:91.00ms
step:588/4740 train_time:53576ms step_avg:91.12ms
step:589/4740 train_time:53674ms step_avg:91.13ms
step:590/4740 train_time:53833ms step_avg:91.24ms
step:591/4740 train_time:53933ms step_avg:91.26ms
step:592/4740 train_time:54092ms step_avg:91.37ms
step:593/4740 train_time:54191ms step_avg:91.38ms
step:594/4740 train_time:54351ms step_avg:91.50ms
step:595/4740 train_time:54450ms step_avg:91.51ms
step:596/4740 train_time:54608ms step_avg:91.62ms
step:597/4740 train_time:54707ms step_avg:91.64ms
step:598/4740 train_time:54865ms step_avg:91.75ms
step:599/4740 train_time:54965ms step_avg:91.76ms
step:600/4740 train_time:55123ms step_avg:91.87ms
step:601/4740 train_time:55223ms step_avg:91.88ms
step:602/4740 train_time:55382ms step_avg:92.00ms
step:603/4740 train_time:55482ms step_avg:92.01ms
step:604/4740 train_time:55640ms step_avg:92.12ms
step:605/4740 train_time:55739ms step_avg:92.13ms
step:606/4740 train_time:55899ms step_avg:92.24ms
step:607/4740 train_time:55998ms step_avg:92.25ms
step:608/4740 train_time:56157ms step_avg:92.36ms
step:609/4740 train_time:56257ms step_avg:92.38ms
step:610/4740 train_time:56417ms step_avg:92.49ms
step:611/4740 train_time:56516ms step_avg:92.50ms
step:612/4740 train_time:56675ms step_avg:92.61ms
step:613/4740 train_time:56774ms step_avg:92.62ms
step:614/4740 train_time:56933ms step_avg:92.73ms
step:615/4740 train_time:57033ms step_avg:92.74ms
step:616/4740 train_time:57193ms step_avg:92.85ms
step:617/4740 train_time:57291ms step_avg:92.85ms
step:618/4740 train_time:57451ms step_avg:92.96ms
step:619/4740 train_time:57550ms step_avg:92.97ms
step:620/4740 train_time:57708ms step_avg:93.08ms
step:621/4740 train_time:57808ms step_avg:93.09ms
step:622/4740 train_time:57965ms step_avg:93.19ms
step:623/4740 train_time:58065ms step_avg:93.20ms
step:624/4740 train_time:58225ms step_avg:93.31ms
step:625/4740 train_time:58324ms step_avg:93.32ms
step:626/4740 train_time:58483ms step_avg:93.42ms
step:627/4740 train_time:58582ms step_avg:93.43ms
step:628/4740 train_time:58742ms step_avg:93.54ms
step:629/4740 train_time:58841ms step_avg:93.55ms
step:630/4740 train_time:59001ms step_avg:93.65ms
step:631/4740 train_time:59101ms step_avg:93.66ms
step:632/4740 train_time:59260ms step_avg:93.77ms
step:633/4740 train_time:59358ms step_avg:93.77ms
step:634/4740 train_time:59517ms step_avg:93.88ms
step:635/4740 train_time:59617ms step_avg:93.89ms
step:636/4740 train_time:59776ms step_avg:93.99ms
step:637/4740 train_time:59877ms step_avg:94.00ms
step:638/4740 train_time:60036ms step_avg:94.10ms
step:639/4740 train_time:60136ms step_avg:94.11ms
step:640/4740 train_time:60295ms step_avg:94.21ms
step:641/4740 train_time:60394ms step_avg:94.22ms
step:642/4740 train_time:60552ms step_avg:94.32ms
step:643/4740 train_time:60651ms step_avg:94.32ms
step:644/4740 train_time:60809ms step_avg:94.42ms
step:645/4740 train_time:60909ms step_avg:94.43ms
step:646/4740 train_time:61067ms step_avg:94.53ms
step:647/4740 train_time:61167ms step_avg:94.54ms
step:648/4740 train_time:61325ms step_avg:94.64ms
step:649/4740 train_time:61426ms step_avg:94.65ms
step:650/4740 train_time:61584ms step_avg:94.74ms
step:651/4740 train_time:61683ms step_avg:94.75ms
step:652/4740 train_time:61842ms step_avg:94.85ms
step:653/4740 train_time:61942ms step_avg:94.86ms
step:654/4740 train_time:62101ms step_avg:94.96ms
step:655/4740 train_time:62202ms step_avg:94.96ms
step:656/4740 train_time:62361ms step_avg:95.06ms
step:657/4740 train_time:62461ms step_avg:95.07ms
step:658/4740 train_time:62620ms step_avg:95.17ms
step:659/4740 train_time:62722ms step_avg:95.18ms
step:660/4740 train_time:62880ms step_avg:95.27ms
step:661/4740 train_time:62980ms step_avg:95.28ms
step:662/4740 train_time:63139ms step_avg:95.38ms
step:663/4740 train_time:63239ms step_avg:95.38ms
step:664/4740 train_time:63398ms step_avg:95.48ms
step:665/4740 train_time:63498ms step_avg:95.49ms
step:666/4740 train_time:63658ms step_avg:95.58ms
step:667/4740 train_time:63757ms step_avg:95.59ms
step:668/4740 train_time:63915ms step_avg:95.68ms
step:669/4740 train_time:64016ms step_avg:95.69ms
step:670/4740 train_time:64175ms step_avg:95.78ms
step:671/4740 train_time:64273ms step_avg:95.79ms
step:672/4740 train_time:64432ms step_avg:95.88ms
step:673/4740 train_time:64531ms step_avg:95.89ms
step:674/4740 train_time:64690ms step_avg:95.98ms
step:675/4740 train_time:64789ms step_avg:95.98ms
step:676/4740 train_time:64948ms step_avg:96.08ms
step:677/4740 train_time:65047ms step_avg:96.08ms
step:678/4740 train_time:65206ms step_avg:96.17ms
step:679/4740 train_time:65304ms step_avg:96.18ms
step:680/4740 train_time:65463ms step_avg:96.27ms
step:681/4740 train_time:65562ms step_avg:96.27ms
step:682/4740 train_time:65722ms step_avg:96.37ms
step:683/4740 train_time:65821ms step_avg:96.37ms
step:684/4740 train_time:65979ms step_avg:96.46ms
step:685/4740 train_time:66079ms step_avg:96.47ms
step:686/4740 train_time:66238ms step_avg:96.56ms
step:687/4740 train_time:66338ms step_avg:96.56ms
step:688/4740 train_time:66496ms step_avg:96.65ms
step:689/4740 train_time:66596ms step_avg:96.66ms
step:690/4740 train_time:66755ms step_avg:96.75ms
step:691/4740 train_time:66853ms step_avg:96.75ms
step:692/4740 train_time:67012ms step_avg:96.84ms
step:693/4740 train_time:67112ms step_avg:96.84ms
step:694/4740 train_time:67271ms step_avg:96.93ms
step:695/4740 train_time:67371ms step_avg:96.94ms
step:696/4740 train_time:67530ms step_avg:97.03ms
step:697/4740 train_time:67629ms step_avg:97.03ms
step:698/4740 train_time:67787ms step_avg:97.12ms
step:699/4740 train_time:67886ms step_avg:97.12ms
step:700/4740 train_time:68045ms step_avg:97.21ms
step:701/4740 train_time:68144ms step_avg:97.21ms
step:702/4740 train_time:68303ms step_avg:97.30ms
step:703/4740 train_time:68403ms step_avg:97.30ms
step:704/4740 train_time:68562ms step_avg:97.39ms
step:705/4740 train_time:68662ms step_avg:97.39ms
step:706/4740 train_time:68821ms step_avg:97.48ms
step:707/4740 train_time:68922ms step_avg:97.48ms
step:708/4740 train_time:69079ms step_avg:97.57ms
step:709/4740 train_time:69179ms step_avg:97.57ms
step:710/4740 train_time:69339ms step_avg:97.66ms
step:711/4740 train_time:69440ms step_avg:97.66ms
step:712/4740 train_time:69597ms step_avg:97.75ms
step:713/4740 train_time:69699ms step_avg:97.76ms
step:714/4740 train_time:69858ms step_avg:97.84ms
step:715/4740 train_time:69959ms step_avg:97.84ms
step:716/4740 train_time:70118ms step_avg:97.93ms
step:717/4740 train_time:70219ms step_avg:97.93ms
step:718/4740 train_time:70377ms step_avg:98.02ms
step:719/4740 train_time:70476ms step_avg:98.02ms
step:720/4740 train_time:70635ms step_avg:98.10ms
step:721/4740 train_time:70734ms step_avg:98.10ms
step:722/4740 train_time:70893ms step_avg:98.19ms
step:723/4740 train_time:70992ms step_avg:98.19ms
step:724/4740 train_time:71152ms step_avg:98.28ms
step:725/4740 train_time:71251ms step_avg:98.28ms
step:726/4740 train_time:71411ms step_avg:98.36ms
step:727/4740 train_time:71510ms step_avg:98.36ms
step:728/4740 train_time:71668ms step_avg:98.45ms
step:729/4740 train_time:71768ms step_avg:98.45ms
step:730/4740 train_time:71926ms step_avg:98.53ms
step:731/4740 train_time:72025ms step_avg:98.53ms
step:732/4740 train_time:72185ms step_avg:98.61ms
step:733/4740 train_time:72284ms step_avg:98.61ms
step:734/4740 train_time:72445ms step_avg:98.70ms
step:735/4740 train_time:72544ms step_avg:98.70ms
step:736/4740 train_time:72704ms step_avg:98.78ms
step:737/4740 train_time:72803ms step_avg:98.78ms
step:738/4740 train_time:72962ms step_avg:98.86ms
step:739/4740 train_time:73062ms step_avg:98.87ms
step:740/4740 train_time:73222ms step_avg:98.95ms
step:741/4740 train_time:73322ms step_avg:98.95ms
step:742/4740 train_time:73481ms step_avg:99.03ms
step:743/4740 train_time:73581ms step_avg:99.03ms
step:744/4740 train_time:73740ms step_avg:99.11ms
step:745/4740 train_time:73840ms step_avg:99.11ms
step:746/4740 train_time:73998ms step_avg:99.19ms
step:747/4740 train_time:74099ms step_avg:99.20ms
step:748/4740 train_time:74258ms step_avg:99.28ms
step:749/4740 train_time:74358ms step_avg:99.28ms
step:750/4740 train_time:74518ms step_avg:99.36ms
step:750/4740 val_loss:3.6401 train_time:74591ms step_avg:99.45ms
step:751/4740 train_time:74620ms step_avg:99.36ms
step:752/4740 train_time:74778ms step_avg:99.44ms
step:753/4740 train_time:74880ms step_avg:99.44ms
step:754/4740 train_time:75047ms step_avg:99.53ms
step:755/4740 train_time:75146ms step_avg:99.53ms
step:756/4740 train_time:75302ms step_avg:99.61ms
step:757/4740 train_time:75400ms step_avg:99.60ms
step:758/4740 train_time:75557ms step_avg:99.68ms
step:759/4740 train_time:75654ms step_avg:99.68ms
step:760/4740 train_time:75813ms step_avg:99.75ms
step:761/4740 train_time:75916ms step_avg:99.76ms
step:762/4740 train_time:76079ms step_avg:99.84ms
step:763/4740 train_time:76179ms step_avg:99.84ms
step:764/4740 train_time:76337ms step_avg:99.92ms
step:765/4740 train_time:76434ms step_avg:99.91ms
step:766/4740 train_time:76591ms step_avg:99.99ms
step:767/4740 train_time:76690ms step_avg:99.99ms
step:768/4740 train_time:76851ms step_avg:100.07ms
step:769/4740 train_time:76953ms step_avg:100.07ms
step:770/4740 train_time:77112ms step_avg:100.15ms
step:771/4740 train_time:77212ms step_avg:100.14ms
step:772/4740 train_time:77370ms step_avg:100.22ms
step:773/4740 train_time:77469ms step_avg:100.22ms
step:774/4740 train_time:77626ms step_avg:100.29ms
step:775/4740 train_time:77724ms step_avg:100.29ms
step:776/4740 train_time:77886ms step_avg:100.37ms
step:777/4740 train_time:77985ms step_avg:100.37ms
step:778/4740 train_time:78145ms step_avg:100.44ms
step:779/4740 train_time:78244ms step_avg:100.44ms
step:780/4740 train_time:78403ms step_avg:100.52ms
step:781/4740 train_time:78501ms step_avg:100.51ms
step:782/4740 train_time:78659ms step_avg:100.59ms
step:783/4740 train_time:78759ms step_avg:100.59ms
step:784/4740 train_time:78919ms step_avg:100.66ms
step:785/4740 train_time:79031ms step_avg:100.68ms
step:786/4740 train_time:79267ms step_avg:100.85ms
step:787/4740 train_time:79409ms step_avg:100.90ms
step:788/4740 train_time:79648ms step_avg:101.08ms
step:789/4740 train_time:79786ms step_avg:101.12ms
step:790/4740 train_time:80028ms step_avg:101.30ms
step:791/4740 train_time:80167ms step_avg:101.35ms
step:792/4740 train_time:80409ms step_avg:101.53ms
step:793/4740 train_time:80546ms step_avg:101.57ms
step:794/4740 train_time:80788ms step_avg:101.75ms
step:795/4740 train_time:80926ms step_avg:101.79ms
step:796/4740 train_time:81169ms step_avg:101.97ms
step:797/4740 train_time:81308ms step_avg:102.02ms
step:798/4740 train_time:81550ms step_avg:102.19ms
step:799/4740 train_time:81687ms step_avg:102.24ms
step:800/4740 train_time:81930ms step_avg:102.41ms
step:801/4740 train_time:82066ms step_avg:102.45ms
step:802/4740 train_time:82308ms step_avg:102.63ms
step:803/4740 train_time:82446ms step_avg:102.67ms
step:804/4740 train_time:82688ms step_avg:102.85ms
step:805/4740 train_time:82826ms step_avg:102.89ms
step:806/4740 train_time:83069ms step_avg:103.06ms
step:807/4740 train_time:83207ms step_avg:103.11ms
step:808/4740 train_time:83449ms step_avg:103.28ms
step:809/4740 train_time:83585ms step_avg:103.32ms
step:810/4740 train_time:83827ms step_avg:103.49ms
step:811/4740 train_time:83963ms step_avg:103.53ms
step:812/4740 train_time:84206ms step_avg:103.70ms
step:813/4740 train_time:84342ms step_avg:103.74ms
step:814/4740 train_time:84586ms step_avg:103.91ms
step:815/4740 train_time:84721ms step_avg:103.95ms
step:816/4740 train_time:84963ms step_avg:104.12ms
step:817/4740 train_time:85099ms step_avg:104.16ms
step:818/4740 train_time:85338ms step_avg:104.33ms
step:819/4740 train_time:85476ms step_avg:104.37ms
step:820/4740 train_time:85717ms step_avg:104.53ms
step:821/4740 train_time:85854ms step_avg:104.57ms
step:822/4740 train_time:86095ms step_avg:104.74ms
step:823/4740 train_time:86232ms step_avg:104.78ms
step:824/4740 train_time:86476ms step_avg:104.95ms
step:825/4740 train_time:86613ms step_avg:104.99ms
step:826/4740 train_time:86855ms step_avg:105.15ms
step:827/4740 train_time:86989ms step_avg:105.19ms
step:828/4740 train_time:87232ms step_avg:105.35ms
step:829/4740 train_time:87369ms step_avg:105.39ms
step:830/4740 train_time:87612ms step_avg:105.56ms
step:831/4740 train_time:87747ms step_avg:105.59ms
step:832/4740 train_time:87988ms step_avg:105.76ms
step:833/4740 train_time:88126ms step_avg:105.79ms
step:834/4740 train_time:88367ms step_avg:105.96ms
step:835/4740 train_time:88506ms step_avg:106.00ms
step:836/4740 train_time:88748ms step_avg:106.16ms
step:837/4740 train_time:88887ms step_avg:106.20ms
step:838/4740 train_time:89130ms step_avg:106.36ms
step:839/4740 train_time:89265ms step_avg:106.39ms
step:840/4740 train_time:89509ms step_avg:106.56ms
step:841/4740 train_time:89647ms step_avg:106.60ms
step:842/4740 train_time:89889ms step_avg:106.76ms
step:843/4740 train_time:90027ms step_avg:106.79ms
step:844/4740 train_time:90267ms step_avg:106.95ms
step:845/4740 train_time:90407ms step_avg:106.99ms
step:846/4740 train_time:90649ms step_avg:107.15ms
step:847/4740 train_time:90785ms step_avg:107.18ms
step:848/4740 train_time:91027ms step_avg:107.34ms
step:849/4740 train_time:91162ms step_avg:107.38ms
step:850/4740 train_time:91408ms step_avg:107.54ms
step:851/4740 train_time:91544ms step_avg:107.57ms
step:852/4740 train_time:91788ms step_avg:107.73ms
step:853/4740 train_time:91922ms step_avg:107.76ms
step:854/4740 train_time:92163ms step_avg:107.92ms
step:855/4740 train_time:92300ms step_avg:107.95ms
step:856/4740 train_time:92542ms step_avg:108.11ms
step:857/4740 train_time:92677ms step_avg:108.14ms
step:858/4740 train_time:92918ms step_avg:108.30ms
step:859/4740 train_time:93055ms step_avg:108.33ms
step:860/4740 train_time:93298ms step_avg:108.49ms
step:861/4740 train_time:93436ms step_avg:108.52ms
step:862/4740 train_time:93675ms step_avg:108.67ms
step:863/4740 train_time:93812ms step_avg:108.70ms
step:864/4740 train_time:94053ms step_avg:108.86ms
step:865/4740 train_time:94190ms step_avg:108.89ms
step:866/4740 train_time:94433ms step_avg:109.04ms
step:867/4740 train_time:94572ms step_avg:109.08ms
step:868/4740 train_time:94811ms step_avg:109.23ms
step:869/4740 train_time:94947ms step_avg:109.26ms
step:870/4740 train_time:95188ms step_avg:109.41ms
step:871/4740 train_time:95326ms step_avg:109.44ms
step:872/4740 train_time:95567ms step_avg:109.60ms
step:873/4740 train_time:95705ms step_avg:109.63ms
step:874/4740 train_time:95945ms step_avg:109.78ms
step:875/4740 train_time:96084ms step_avg:109.81ms
step:876/4740 train_time:96330ms step_avg:109.97ms
step:877/4740 train_time:96464ms step_avg:109.99ms
step:878/4740 train_time:96707ms step_avg:110.14ms
step:879/4740 train_time:96840ms step_avg:110.17ms
step:880/4740 train_time:97081ms step_avg:110.32ms
step:881/4740 train_time:97217ms step_avg:110.35ms
step:882/4740 train_time:97457ms step_avg:110.49ms
step:883/4740 train_time:97593ms step_avg:110.52ms
step:884/4740 train_time:97832ms step_avg:110.67ms
step:885/4740 train_time:97971ms step_avg:110.70ms
step:886/4740 train_time:98211ms step_avg:110.85ms
step:887/4740 train_time:98347ms step_avg:110.88ms
step:888/4740 train_time:98590ms step_avg:111.03ms
step:889/4740 train_time:98726ms step_avg:111.05ms
step:890/4740 train_time:98969ms step_avg:111.20ms
step:891/4740 train_time:99104ms step_avg:111.23ms
step:892/4740 train_time:99346ms step_avg:111.37ms
step:893/4740 train_time:99484ms step_avg:111.40ms
step:894/4740 train_time:99727ms step_avg:111.55ms
step:895/4740 train_time:99863ms step_avg:111.58ms
step:896/4740 train_time:100105ms step_avg:111.72ms
step:897/4740 train_time:100240ms step_avg:111.75ms
step:898/4740 train_time:100482ms step_avg:111.89ms
step:899/4740 train_time:100618ms step_avg:111.92ms
step:900/4740 train_time:100858ms step_avg:112.06ms
step:901/4740 train_time:100993ms step_avg:112.09ms
step:902/4740 train_time:101232ms step_avg:112.23ms
step:903/4740 train_time:101369ms step_avg:112.26ms
step:904/4740 train_time:101615ms step_avg:112.41ms
step:905/4740 train_time:101752ms step_avg:112.43ms
step:906/4740 train_time:101989ms step_avg:112.57ms
step:907/4740 train_time:102127ms step_avg:112.60ms
step:908/4740 train_time:102369ms step_avg:112.74ms
step:909/4740 train_time:102507ms step_avg:112.77ms
step:910/4740 train_time:102749ms step_avg:112.91ms
step:911/4740 train_time:102886ms step_avg:112.94ms
step:912/4740 train_time:103129ms step_avg:113.08ms
step:913/4740 train_time:103268ms step_avg:113.11ms
step:914/4740 train_time:103508ms step_avg:113.25ms
step:915/4740 train_time:103645ms step_avg:113.27ms
step:916/4740 train_time:103887ms step_avg:113.41ms
step:917/4740 train_time:104022ms step_avg:113.44ms
step:918/4740 train_time:104265ms step_avg:113.58ms
step:919/4740 train_time:104401ms step_avg:113.60ms
step:920/4740 train_time:104644ms step_avg:113.74ms
step:921/4740 train_time:104780ms step_avg:113.77ms
step:922/4740 train_time:105019ms step_avg:113.90ms
step:923/4740 train_time:105156ms step_avg:113.93ms
step:924/4740 train_time:105395ms step_avg:114.06ms
step:925/4740 train_time:105533ms step_avg:114.09ms
step:926/4740 train_time:105774ms step_avg:114.23ms
step:927/4740 train_time:105911ms step_avg:114.25ms
step:928/4740 train_time:106153ms step_avg:114.39ms
step:929/4740 train_time:106289ms step_avg:114.41ms
step:930/4740 train_time:106534ms step_avg:114.55ms
step:931/4740 train_time:106669ms step_avg:114.57ms
step:932/4740 train_time:106910ms step_avg:114.71ms
step:933/4740 train_time:107047ms step_avg:114.73ms
step:934/4740 train_time:107290ms step_avg:114.87ms
step:935/4740 train_time:107425ms step_avg:114.89ms
step:936/4740 train_time:107666ms step_avg:115.03ms
step:937/4740 train_time:107805ms step_avg:115.05ms
step:938/4740 train_time:108048ms step_avg:115.19ms
step:939/4740 train_time:108183ms step_avg:115.21ms
step:940/4740 train_time:108425ms step_avg:115.35ms
step:941/4740 train_time:108561ms step_avg:115.37ms
step:942/4740 train_time:108804ms step_avg:115.50ms
step:943/4740 train_time:108940ms step_avg:115.52ms
step:944/4740 train_time:109178ms step_avg:115.65ms
step:945/4740 train_time:109315ms step_avg:115.68ms
step:946/4740 train_time:109553ms step_avg:115.81ms
step:947/4740 train_time:109692ms step_avg:115.83ms
step:948/4740 train_time:109934ms step_avg:115.96ms
step:949/4740 train_time:110069ms step_avg:115.98ms
step:950/4740 train_time:110308ms step_avg:116.11ms
step:951/4740 train_time:110446ms step_avg:116.14ms
step:952/4740 train_time:110689ms step_avg:116.27ms
step:953/4740 train_time:110825ms step_avg:116.29ms
step:954/4740 train_time:111066ms step_avg:116.42ms
step:955/4740 train_time:111202ms step_avg:116.44ms
step:956/4740 train_time:111445ms step_avg:116.57ms
step:957/4740 train_time:111580ms step_avg:116.59ms
step:958/4740 train_time:111823ms step_avg:116.73ms
step:959/4740 train_time:111959ms step_avg:116.75ms
step:960/4740 train_time:112198ms step_avg:116.87ms
step:961/4740 train_time:112337ms step_avg:116.90ms
step:962/4740 train_time:112576ms step_avg:117.02ms
step:963/4740 train_time:112715ms step_avg:117.05ms
step:964/4740 train_time:112955ms step_avg:117.17ms
step:965/4740 train_time:113092ms step_avg:117.19ms
step:966/4740 train_time:113332ms step_avg:117.32ms
step:967/4740 train_time:113470ms step_avg:117.34ms
step:968/4740 train_time:113713ms step_avg:117.47ms
step:969/4740 train_time:113849ms step_avg:117.49ms
step:970/4740 train_time:114091ms step_avg:117.62ms
step:971/4740 train_time:114226ms step_avg:117.64ms
step:972/4740 train_time:114469ms step_avg:117.77ms
step:973/4740 train_time:114610ms step_avg:117.79ms
step:974/4740 train_time:114851ms step_avg:117.92ms
step:975/4740 train_time:114989ms step_avg:117.94ms
step:976/4740 train_time:115232ms step_avg:118.07ms
step:977/4740 train_time:115371ms step_avg:118.09ms
step:978/4740 train_time:115612ms step_avg:118.21ms
step:979/4740 train_time:115747ms step_avg:118.23ms
step:980/4740 train_time:115990ms step_avg:118.36ms
step:981/4740 train_time:116127ms step_avg:118.38ms
step:982/4740 train_time:116369ms step_avg:118.50ms
step:983/4740 train_time:116506ms step_avg:118.52ms
step:984/4740 train_time:116747ms step_avg:118.64ms
step:985/4740 train_time:116884ms step_avg:118.66ms
step:986/4740 train_time:117126ms step_avg:118.79ms
step:987/4740 train_time:117261ms step_avg:118.81ms
step:988/4740 train_time:117505ms step_avg:118.93ms
step:989/4740 train_time:117641ms step_avg:118.95ms
step:990/4740 train_time:117883ms step_avg:119.07ms
step:991/4740 train_time:118019ms step_avg:119.09ms
step:992/4740 train_time:118261ms step_avg:119.21ms
step:993/4740 train_time:118397ms step_avg:119.23ms
step:994/4740 train_time:118636ms step_avg:119.35ms
step:995/4740 train_time:118774ms step_avg:119.37ms
step:996/4740 train_time:119015ms step_avg:119.49ms
step:997/4740 train_time:119153ms step_avg:119.51ms
step:998/4740 train_time:119394ms step_avg:119.63ms
step:999/4740 train_time:119531ms step_avg:119.65ms
step:1000/4740 train_time:119770ms step_avg:119.77ms
step:1000/4740 val_loss:3.4782 train_time:119868ms step_avg:119.87ms
step:1001/4740 train_time:119910ms step_avg:119.79ms
step:1002/4740 train_time:120151ms step_avg:119.91ms
step:1003/4740 train_time:120298ms step_avg:119.94ms
step:1004/4740 train_time:120541ms step_avg:120.06ms
step:1005/4740 train_time:120675ms step_avg:120.07ms
step:1006/4740 train_time:120913ms step_avg:120.19ms
step:1007/4740 train_time:121049ms step_avg:120.21ms
step:1008/4740 train_time:121293ms step_avg:120.33ms
step:1009/4740 train_time:121433ms step_avg:120.35ms
step:1010/4740 train_time:121672ms step_avg:120.47ms
step:1011/4740 train_time:121808ms step_avg:120.48ms
step:1012/4740 train_time:122046ms step_avg:120.60ms
step:1013/4740 train_time:122184ms step_avg:120.62ms
step:1014/4740 train_time:122430ms step_avg:120.74ms
step:1015/4740 train_time:122567ms step_avg:120.76ms
step:1016/4740 train_time:122805ms step_avg:120.87ms
step:1017/4740 train_time:122940ms step_avg:120.88ms
step:1018/4740 train_time:123180ms step_avg:121.00ms
step:1019/4740 train_time:123321ms step_avg:121.02ms
step:1020/4740 train_time:123564ms step_avg:121.14ms
step:1021/4740 train_time:123700ms step_avg:121.16ms
step:1022/4740 train_time:123941ms step_avg:121.27ms
step:1023/4740 train_time:124078ms step_avg:121.29ms
step:1024/4740 train_time:124324ms step_avg:121.41ms
step:1025/4740 train_time:124460ms step_avg:121.42ms
step:1026/4740 train_time:124703ms step_avg:121.54ms
step:1027/4740 train_time:124838ms step_avg:121.56ms
step:1028/4740 train_time:125081ms step_avg:121.67ms
step:1029/4740 train_time:125220ms step_avg:121.69ms
step:1030/4740 train_time:125463ms step_avg:121.81ms
step:1031/4740 train_time:125600ms step_avg:121.82ms
step:1032/4740 train_time:125839ms step_avg:121.94ms
step:1033/4740 train_time:125979ms step_avg:121.95ms
step:1034/4740 train_time:126218ms step_avg:122.07ms
step:1035/4740 train_time:126359ms step_avg:122.09ms
step:1036/4740 train_time:126600ms step_avg:122.20ms
step:1037/4740 train_time:126739ms step_avg:122.22ms
step:1038/4740 train_time:126981ms step_avg:122.33ms
step:1039/4740 train_time:127117ms step_avg:122.35ms
step:1040/4740 train_time:127361ms step_avg:122.46ms
step:1041/4740 train_time:127497ms step_avg:122.48ms
step:1042/4740 train_time:127739ms step_avg:122.59ms
step:1043/4740 train_time:127878ms step_avg:122.61ms
step:1044/4740 train_time:128120ms step_avg:122.72ms
step:1045/4740 train_time:128257ms step_avg:122.73ms
step:1046/4740 train_time:128500ms step_avg:122.85ms
step:1047/4740 train_time:128637ms step_avg:122.86ms
step:1048/4740 train_time:128879ms step_avg:122.98ms
step:1049/4740 train_time:129014ms step_avg:122.99ms
step:1050/4740 train_time:129259ms step_avg:123.10ms
step:1051/4740 train_time:129397ms step_avg:123.12ms
step:1052/4740 train_time:129640ms step_avg:123.23ms
step:1053/4740 train_time:129776ms step_avg:123.24ms
step:1054/4740 train_time:130019ms step_avg:123.36ms
step:1055/4740 train_time:130152ms step_avg:123.37ms
step:1056/4740 train_time:130396ms step_avg:123.48ms
step:1057/4740 train_time:130531ms step_avg:123.49ms
step:1058/4740 train_time:130774ms step_avg:123.60ms
step:1059/4740 train_time:130909ms step_avg:123.62ms
step:1060/4740 train_time:131148ms step_avg:123.72ms
step:1061/4740 train_time:131287ms step_avg:123.74ms
step:1062/4740 train_time:131525ms step_avg:123.85ms
step:1063/4740 train_time:131663ms step_avg:123.86ms
step:1064/4740 train_time:131903ms step_avg:123.97ms
step:1065/4740 train_time:132041ms step_avg:123.98ms
step:1066/4740 train_time:132281ms step_avg:124.09ms
step:1067/4740 train_time:132419ms step_avg:124.10ms
step:1068/4740 train_time:132661ms step_avg:124.21ms
step:1069/4740 train_time:132798ms step_avg:124.23ms
step:1070/4740 train_time:133040ms step_avg:124.34ms
step:1071/4740 train_time:133177ms step_avg:124.35ms
step:1072/4740 train_time:133419ms step_avg:124.46ms
step:1073/4740 train_time:133556ms step_avg:124.47ms
step:1074/4740 train_time:133798ms step_avg:124.58ms
step:1075/4740 train_time:133936ms step_avg:124.59ms
step:1076/4740 train_time:134178ms step_avg:124.70ms
step:1077/4740 train_time:134312ms step_avg:124.71ms
step:1078/4740 train_time:134555ms step_avg:124.82ms
step:1079/4740 train_time:134690ms step_avg:124.83ms
step:1080/4740 train_time:134928ms step_avg:124.93ms
step:1081/4740 train_time:135065ms step_avg:124.94ms
step:1082/4740 train_time:135308ms step_avg:125.05ms
step:1083/4740 train_time:135444ms step_avg:125.06ms
step:1084/4740 train_time:135683ms step_avg:125.17ms
step:1085/4740 train_time:135820ms step_avg:125.18ms
step:1086/4740 train_time:136061ms step_avg:125.29ms
step:1087/4740 train_time:136198ms step_avg:125.30ms
step:1088/4740 train_time:136440ms step_avg:125.40ms
step:1089/4740 train_time:136575ms step_avg:125.41ms
step:1090/4740 train_time:136819ms step_avg:125.52ms
step:1091/4740 train_time:136954ms step_avg:125.53ms
step:1092/4740 train_time:137199ms step_avg:125.64ms
step:1093/4740 train_time:137333ms step_avg:125.65ms
step:1094/4740 train_time:137573ms step_avg:125.75ms
step:1095/4740 train_time:137710ms step_avg:125.76ms
step:1096/4740 train_time:137949ms step_avg:125.87ms
step:1097/4740 train_time:138087ms step_avg:125.88ms
step:1098/4740 train_time:138327ms step_avg:125.98ms
step:1099/4740 train_time:138464ms step_avg:125.99ms
step:1100/4740 train_time:138707ms step_avg:126.10ms
step:1101/4740 train_time:138847ms step_avg:126.11ms
step:1102/4740 train_time:139090ms step_avg:126.22ms
step:1103/4740 train_time:139228ms step_avg:126.23ms
step:1104/4740 train_time:139467ms step_avg:126.33ms
step:1105/4740 train_time:139604ms step_avg:126.34ms
step:1106/4740 train_time:139846ms step_avg:126.44ms
step:1107/4740 train_time:139984ms step_avg:126.45ms
step:1108/4740 train_time:140225ms step_avg:126.56ms
step:1109/4740 train_time:140364ms step_avg:126.57ms
step:1110/4740 train_time:140603ms step_avg:126.67ms
step:1111/4740 train_time:140741ms step_avg:126.68ms
step:1112/4740 train_time:140983ms step_avg:126.78ms
step:1113/4740 train_time:141121ms step_avg:126.79ms
step:1114/4740 train_time:141361ms step_avg:126.90ms
step:1115/4740 train_time:141498ms step_avg:126.90ms
step:1116/4740 train_time:141740ms step_avg:127.01ms
step:1117/4740 train_time:141877ms step_avg:127.02ms
step:1118/4740 train_time:142120ms step_avg:127.12ms
step:1119/4740 train_time:142255ms step_avg:127.13ms
step:1120/4740 train_time:142497ms step_avg:127.23ms
step:1121/4740 train_time:142636ms step_avg:127.24ms
step:1122/4740 train_time:142877ms step_avg:127.34ms
step:1123/4740 train_time:143015ms step_avg:127.35ms
step:1124/4740 train_time:143255ms step_avg:127.45ms
step:1125/4740 train_time:143392ms step_avg:127.46ms
step:1126/4740 train_time:143631ms step_avg:127.56ms
step:1127/4740 train_time:143770ms step_avg:127.57ms
step:1128/4740 train_time:144010ms step_avg:127.67ms
step:1129/4740 train_time:144147ms step_avg:127.68ms
step:1130/4740 train_time:144389ms step_avg:127.78ms
step:1131/4740 train_time:144525ms step_avg:127.79ms
step:1132/4740 train_time:144765ms step_avg:127.88ms
step:1133/4740 train_time:144904ms step_avg:127.89ms
step:1134/4740 train_time:145147ms step_avg:128.00ms
step:1135/4740 train_time:145283ms step_avg:128.00ms
step:1136/4740 train_time:145522ms step_avg:128.10ms
step:1137/4740 train_time:145661ms step_avg:128.11ms
step:1138/4740 train_time:145904ms step_avg:128.21ms
step:1139/4740 train_time:146042ms step_avg:128.22ms
step:1140/4740 train_time:146283ms step_avg:128.32ms
step:1141/4740 train_time:146419ms step_avg:128.33ms
step:1142/4740 train_time:146662ms step_avg:128.43ms
step:1143/4740 train_time:146799ms step_avg:128.43ms
step:1144/4740 train_time:147040ms step_avg:128.53ms
step:1145/4740 train_time:147179ms step_avg:128.54ms
step:1146/4740 train_time:147421ms step_avg:128.64ms
step:1147/4740 train_time:147560ms step_avg:128.65ms
step:1148/4740 train_time:147798ms step_avg:128.74ms
step:1149/4740 train_time:147940ms step_avg:128.76ms
step:1150/4740 train_time:148182ms step_avg:128.85ms
step:1151/4740 train_time:148320ms step_avg:128.86ms
step:1152/4740 train_time:148559ms step_avg:128.96ms
step:1153/4740 train_time:148696ms step_avg:128.96ms
step:1154/4740 train_time:148939ms step_avg:129.06ms
step:1155/4740 train_time:149079ms step_avg:129.07ms
step:1156/4740 train_time:149318ms step_avg:129.17ms
step:1157/4740 train_time:149455ms step_avg:129.17ms
step:1158/4740 train_time:149697ms step_avg:129.27ms
step:1159/4740 train_time:149834ms step_avg:129.28ms
step:1160/4740 train_time:150074ms step_avg:129.37ms
step:1161/4740 train_time:150211ms step_avg:129.38ms
step:1162/4740 train_time:150452ms step_avg:129.48ms
step:1163/4740 train_time:150589ms step_avg:129.48ms
step:1164/4740 train_time:150829ms step_avg:129.58ms
step:1165/4740 train_time:150967ms step_avg:129.59ms
step:1166/4740 train_time:151209ms step_avg:129.68ms
step:1167/4740 train_time:151345ms step_avg:129.69ms
step:1168/4740 train_time:151588ms step_avg:129.78ms
step:1169/4740 train_time:151722ms step_avg:129.79ms
step:1170/4740 train_time:151965ms step_avg:129.88ms
step:1171/4740 train_time:152100ms step_avg:129.89ms
step:1172/4740 train_time:152344ms step_avg:129.99ms
step:1173/4740 train_time:152479ms step_avg:129.99ms
step:1174/4740 train_time:152722ms step_avg:130.09ms
step:1175/4740 train_time:152857ms step_avg:130.09ms
step:1176/4740 train_time:153132ms step_avg:130.21ms
step:1177/4740 train_time:153308ms step_avg:130.25ms
step:1178/4740 train_time:153623ms step_avg:130.41ms
step:1179/4740 train_time:153800ms step_avg:130.45ms
step:1180/4740 train_time:154116ms step_avg:130.61ms
step:1181/4740 train_time:154292ms step_avg:130.65ms
step:1182/4740 train_time:154609ms step_avg:130.80ms
step:1183/4740 train_time:154787ms step_avg:130.84ms
step:1184/4740 train_time:155102ms step_avg:131.00ms
step:1185/4740 train_time:155278ms step_avg:131.04ms
step:1186/4740 train_time:155598ms step_avg:131.20ms
step:1187/4740 train_time:155774ms step_avg:131.23ms
step:1188/4740 train_time:156095ms step_avg:131.39ms
step:1189/4740 train_time:156269ms step_avg:131.43ms
step:1190/4740 train_time:156591ms step_avg:131.59ms
step:1191/4740 train_time:156765ms step_avg:131.63ms
step:1192/4740 train_time:157082ms step_avg:131.78ms
step:1193/4740 train_time:157259ms step_avg:131.82ms
step:1194/4740 train_time:157578ms step_avg:131.97ms
step:1195/4740 train_time:157753ms step_avg:132.01ms
step:1196/4740 train_time:158074ms step_avg:132.17ms
step:1197/4740 train_time:158247ms step_avg:132.20ms
step:1198/4740 train_time:158566ms step_avg:132.36ms
step:1199/4740 train_time:158744ms step_avg:132.40ms
step:1200/4740 train_time:159061ms step_avg:132.55ms
step:1201/4740 train_time:159233ms step_avg:132.58ms
step:1202/4740 train_time:159552ms step_avg:132.74ms
step:1203/4740 train_time:159728ms step_avg:132.77ms
step:1204/4740 train_time:160044ms step_avg:132.93ms
step:1205/4740 train_time:160218ms step_avg:132.96ms
step:1206/4740 train_time:160537ms step_avg:133.12ms
step:1207/4740 train_time:160714ms step_avg:133.15ms
step:1208/4740 train_time:161030ms step_avg:133.30ms
step:1209/4740 train_time:161205ms step_avg:133.34ms
step:1210/4740 train_time:161523ms step_avg:133.49ms
step:1211/4740 train_time:161699ms step_avg:133.53ms
step:1212/4740 train_time:162017ms step_avg:133.68ms
step:1213/4740 train_time:162193ms step_avg:133.71ms
step:1214/4740 train_time:162513ms step_avg:133.87ms
step:1215/4740 train_time:162688ms step_avg:133.90ms
step:1216/4740 train_time:163004ms step_avg:134.05ms
step:1217/4740 train_time:163181ms step_avg:134.08ms
step:1218/4740 train_time:163498ms step_avg:134.23ms
step:1219/4740 train_time:163673ms step_avg:134.27ms
step:1220/4740 train_time:163995ms step_avg:134.42ms
step:1221/4740 train_time:164170ms step_avg:134.46ms
step:1222/4740 train_time:164487ms step_avg:134.61ms
step:1223/4740 train_time:164662ms step_avg:134.64ms
step:1224/4740 train_time:164982ms step_avg:134.79ms
step:1225/4740 train_time:165155ms step_avg:134.82ms
step:1226/4740 train_time:165474ms step_avg:134.97ms
step:1227/4740 train_time:165651ms step_avg:135.00ms
step:1228/4740 train_time:165973ms step_avg:135.16ms
step:1229/4740 train_time:166148ms step_avg:135.19ms
step:1230/4740 train_time:166467ms step_avg:135.34ms
step:1231/4740 train_time:166641ms step_avg:135.37ms
step:1232/4740 train_time:166957ms step_avg:135.52ms
step:1233/4740 train_time:167135ms step_avg:135.55ms
step:1234/4740 train_time:167449ms step_avg:135.70ms
step:1235/4740 train_time:167626ms step_avg:135.73ms
step:1236/4740 train_time:167941ms step_avg:135.87ms
step:1237/4740 train_time:168117ms step_avg:135.91ms
step:1238/4740 train_time:168433ms step_avg:136.05ms
step:1239/4740 train_time:168611ms step_avg:136.09ms
step:1240/4740 train_time:168930ms step_avg:136.23ms
step:1241/4740 train_time:169105ms step_avg:136.26ms
step:1242/4740 train_time:169420ms step_avg:136.41ms
step:1243/4740 train_time:169597ms step_avg:136.44ms
step:1244/4740 train_time:169912ms step_avg:136.59ms
step:1245/4740 train_time:170087ms step_avg:136.62ms
step:1246/4740 train_time:170405ms step_avg:136.76ms
step:1247/4740 train_time:170581ms step_avg:136.79ms
step:1248/4740 train_time:170897ms step_avg:136.94ms
step:1249/4740 train_time:171077ms step_avg:136.97ms
step:1250/4740 train_time:171394ms step_avg:137.12ms
step:1250/4740 val_loss:3.3997 train_time:171517ms step_avg:137.21ms
step:1251/4740 train_time:171573ms step_avg:137.15ms
step:1252/4740 train_time:171900ms step_avg:137.30ms
step:1253/4740 train_time:172076ms step_avg:137.33ms
step:1254/4740 train_time:172389ms step_avg:137.47ms
step:1255/4740 train_time:172561ms step_avg:137.50ms
step:1256/4740 train_time:172884ms step_avg:137.65ms
step:1257/4740 train_time:173062ms step_avg:137.68ms
step:1258/4740 train_time:173378ms step_avg:137.82ms
step:1259/4740 train_time:173551ms step_avg:137.85ms
step:1260/4740 train_time:173870ms step_avg:137.99ms
step:1261/4740 train_time:174047ms step_avg:138.02ms
step:1262/4740 train_time:174363ms step_avg:138.16ms
step:1263/4740 train_time:174538ms step_avg:138.19ms
step:1264/4740 train_time:174856ms step_avg:138.34ms
step:1265/4740 train_time:175031ms step_avg:138.36ms
step:1266/4740 train_time:175348ms step_avg:138.51ms
step:1267/4740 train_time:175527ms step_avg:138.54ms
step:1268/4740 train_time:175844ms step_avg:138.68ms
step:1269/4740 train_time:176022ms step_avg:138.71ms
step:1270/4740 train_time:176341ms step_avg:138.85ms
step:1271/4740 train_time:176514ms step_avg:138.88ms
step:1272/4740 train_time:176832ms step_avg:139.02ms
step:1273/4740 train_time:177007ms step_avg:139.05ms
step:1274/4740 train_time:177326ms step_avg:139.19ms
step:1275/4740 train_time:177502ms step_avg:139.22ms
step:1276/4740 train_time:177822ms step_avg:139.36ms
step:1277/4740 train_time:177996ms step_avg:139.39ms
step:1278/4740 train_time:178310ms step_avg:139.52ms
step:1279/4740 train_time:178489ms step_avg:139.55ms
step:1280/4740 train_time:178807ms step_avg:139.69ms
step:1281/4740 train_time:178983ms step_avg:139.72ms
step:1282/4740 train_time:179302ms step_avg:139.86ms
step:1283/4740 train_time:179478ms step_avg:139.89ms
step:1284/4740 train_time:179794ms step_avg:140.03ms
step:1285/4740 train_time:179970ms step_avg:140.05ms
step:1286/4740 train_time:180290ms step_avg:140.19ms
step:1287/4740 train_time:180463ms step_avg:140.22ms
step:1288/4740 train_time:180782ms step_avg:140.36ms
step:1289/4740 train_time:180955ms step_avg:140.38ms
step:1290/4740 train_time:181272ms step_avg:140.52ms
step:1291/4740 train_time:181446ms step_avg:140.55ms
step:1292/4740 train_time:181765ms step_avg:140.68ms
step:1293/4740 train_time:181942ms step_avg:140.71ms
step:1294/4740 train_time:182260ms step_avg:140.85ms
step:1295/4740 train_time:182438ms step_avg:140.88ms
step:1296/4740 train_time:182754ms step_avg:141.01ms
step:1297/4740 train_time:182929ms step_avg:141.04ms
step:1298/4740 train_time:183245ms step_avg:141.17ms
step:1299/4740 train_time:183424ms step_avg:141.20ms
step:1300/4740 train_time:183742ms step_avg:141.34ms
step:1301/4740 train_time:183919ms step_avg:141.37ms
step:1302/4740 train_time:184234ms step_avg:141.50ms
step:1303/4740 train_time:184413ms step_avg:141.53ms
step:1304/4740 train_time:184730ms step_avg:141.66ms
step:1305/4740 train_time:184909ms step_avg:141.69ms
step:1306/4740 train_time:185227ms step_avg:141.83ms
step:1307/4740 train_time:185404ms step_avg:141.85ms
step:1308/4740 train_time:185723ms step_avg:141.99ms
step:1309/4740 train_time:185897ms step_avg:142.01ms
step:1310/4740 train_time:186216ms step_avg:142.15ms
step:1311/4740 train_time:186392ms step_avg:142.18ms
step:1312/4740 train_time:186710ms step_avg:142.31ms
step:1313/4740 train_time:186884ms step_avg:142.33ms
step:1314/4740 train_time:187202ms step_avg:142.47ms
step:1315/4740 train_time:187377ms step_avg:142.49ms
step:1316/4740 train_time:187697ms step_avg:142.63ms
step:1317/4740 train_time:187872ms step_avg:142.65ms
step:1318/4740 train_time:188187ms step_avg:142.78ms
step:1319/4740 train_time:188361ms step_avg:142.81ms
step:1320/4740 train_time:188680ms step_avg:142.94ms
step:1321/4740 train_time:188855ms step_avg:142.96ms
step:1322/4740 train_time:189171ms step_avg:143.09ms
step:1323/4740 train_time:189349ms step_avg:143.12ms
step:1324/4740 train_time:189665ms step_avg:143.25ms
step:1325/4740 train_time:189845ms step_avg:143.28ms
step:1326/4740 train_time:190163ms step_avg:143.41ms
step:1327/4740 train_time:190342ms step_avg:143.44ms
step:1328/4740 train_time:190659ms step_avg:143.57ms
step:1329/4740 train_time:190834ms step_avg:143.59ms
step:1330/4740 train_time:191151ms step_avg:143.72ms
step:1331/4740 train_time:191327ms step_avg:143.75ms
step:1332/4740 train_time:191645ms step_avg:143.88ms
step:1333/4740 train_time:191819ms step_avg:143.90ms
step:1334/4740 train_time:192135ms step_avg:144.03ms
step:1335/4740 train_time:192312ms step_avg:144.05ms
step:1336/4740 train_time:192628ms step_avg:144.18ms
step:1337/4740 train_time:192805ms step_avg:144.21ms
step:1338/4740 train_time:193123ms step_avg:144.34ms
step:1339/4740 train_time:193302ms step_avg:144.36ms
step:1340/4740 train_time:193620ms step_avg:144.49ms
step:1341/4740 train_time:193796ms step_avg:144.52ms
step:1342/4740 train_time:194112ms step_avg:144.64ms
step:1343/4740 train_time:194287ms step_avg:144.67ms
step:1344/4740 train_time:194604ms step_avg:144.79ms
step:1345/4740 train_time:194780ms step_avg:144.82ms
step:1346/4740 train_time:195097ms step_avg:144.95ms
step:1347/4740 train_time:195272ms step_avg:144.97ms
step:1348/4740 train_time:195590ms step_avg:145.10ms
step:1349/4740 train_time:195764ms step_avg:145.12ms
step:1350/4740 train_time:196087ms step_avg:145.25ms
step:1351/4740 train_time:196259ms step_avg:145.27ms
step:1352/4740 train_time:196580ms step_avg:145.40ms
step:1353/4740 train_time:196756ms step_avg:145.42ms
step:1354/4740 train_time:197075ms step_avg:145.55ms
step:1355/4740 train_time:197250ms step_avg:145.57ms
step:1356/4740 train_time:197568ms step_avg:145.70ms
step:1357/4740 train_time:197744ms step_avg:145.72ms
step:1358/4740 train_time:198063ms step_avg:145.85ms
step:1359/4740 train_time:198239ms step_avg:145.87ms
step:1360/4740 train_time:198560ms step_avg:146.00ms
step:1361/4740 train_time:198732ms step_avg:146.02ms
step:1362/4740 train_time:199047ms step_avg:146.14ms
step:1363/4740 train_time:199225ms step_avg:146.17ms
step:1364/4740 train_time:199544ms step_avg:146.29ms
step:1365/4740 train_time:199718ms step_avg:146.31ms
step:1366/4740 train_time:200033ms step_avg:146.44ms
step:1367/4740 train_time:200210ms step_avg:146.46ms
step:1368/4740 train_time:200528ms step_avg:146.58ms
step:1369/4740 train_time:200704ms step_avg:146.61ms
step:1370/4740 train_time:201022ms step_avg:146.73ms
step:1371/4740 train_time:201201ms step_avg:146.76ms
step:1372/4740 train_time:201521ms step_avg:146.88ms
step:1373/4740 train_time:201695ms step_avg:146.90ms
step:1374/4740 train_time:202010ms step_avg:147.02ms
step:1375/4740 train_time:202186ms step_avg:147.04ms
step:1376/4740 train_time:202504ms step_avg:147.17ms
step:1377/4740 train_time:202679ms step_avg:147.19ms
step:1378/4740 train_time:202999ms step_avg:147.31ms
step:1379/4740 train_time:203176ms step_avg:147.34ms
step:1380/4740 train_time:203494ms step_avg:147.46ms
step:1381/4740 train_time:203669ms step_avg:147.48ms
step:1382/4740 train_time:203989ms step_avg:147.60ms
step:1383/4740 train_time:204162ms step_avg:147.62ms
step:1384/4740 train_time:204480ms step_avg:147.75ms
step:1385/4740 train_time:204659ms step_avg:147.77ms
step:1386/4740 train_time:204977ms step_avg:147.89ms
step:1387/4740 train_time:205152ms step_avg:147.91ms
step:1388/4740 train_time:205469ms step_avg:148.03ms
step:1389/4740 train_time:205645ms step_avg:148.05ms
step:1390/4740 train_time:205964ms step_avg:148.18ms
step:1391/4740 train_time:206141ms step_avg:148.20ms
step:1392/4740 train_time:206460ms step_avg:148.32ms
step:1393/4740 train_time:206633ms step_avg:148.34ms
step:1394/4740 train_time:206950ms step_avg:148.46ms
step:1395/4740 train_time:207124ms step_avg:148.48ms
step:1396/4740 train_time:207442ms step_avg:148.60ms
step:1397/4740 train_time:207617ms step_avg:148.62ms
step:1398/4740 train_time:207934ms step_avg:148.74ms
step:1399/4740 train_time:208111ms step_avg:148.76ms
step:1400/4740 train_time:208427ms step_avg:148.88ms
step:1401/4740 train_time:208603ms step_avg:148.90ms
step:1402/4740 train_time:208921ms step_avg:149.02ms
step:1403/4740 train_time:209098ms step_avg:149.04ms
step:1404/4740 train_time:209414ms step_avg:149.16ms
step:1405/4740 train_time:209589ms step_avg:149.17ms
step:1406/4740 train_time:209906ms step_avg:149.29ms
step:1407/4740 train_time:210081ms step_avg:149.31ms
step:1408/4740 train_time:210402ms step_avg:149.43ms
step:1409/4740 train_time:210574ms step_avg:149.45ms
step:1410/4740 train_time:210891ms step_avg:149.57ms
step:1411/4740 train_time:211067ms step_avg:149.59ms
step:1412/4740 train_time:211384ms step_avg:149.71ms
step:1413/4740 train_time:211561ms step_avg:149.72ms
step:1414/4740 train_time:211879ms step_avg:149.84ms
step:1415/4740 train_time:212055ms step_avg:149.86ms
step:1416/4740 train_time:212371ms step_avg:149.98ms
step:1417/4740 train_time:212548ms step_avg:150.00ms
step:1418/4740 train_time:212863ms step_avg:150.11ms
step:1419/4740 train_time:213040ms step_avg:150.13ms
step:1420/4740 train_time:213359ms step_avg:150.25ms
step:1421/4740 train_time:213534ms step_avg:150.27ms
step:1422/4740 train_time:213850ms step_avg:150.39ms
step:1423/4740 train_time:214026ms step_avg:150.40ms
step:1424/4740 train_time:214343ms step_avg:150.52ms
step:1425/4740 train_time:214519ms step_avg:150.54ms
step:1426/4740 train_time:214840ms step_avg:150.66ms
step:1427/4740 train_time:215014ms step_avg:150.68ms
step:1428/4740 train_time:215330ms step_avg:150.79ms
step:1429/4740 train_time:215505ms step_avg:150.81ms
step:1430/4740 train_time:215824ms step_avg:150.93ms
step:1431/4740 train_time:215998ms step_avg:150.94ms
step:1432/4740 train_time:216317ms step_avg:151.06ms
step:1433/4740 train_time:216495ms step_avg:151.08ms
step:1434/4740 train_time:216808ms step_avg:151.19ms
step:1435/4740 train_time:216988ms step_avg:151.21ms
step:1436/4740 train_time:217302ms step_avg:151.32ms
step:1437/4740 train_time:217480ms step_avg:151.34ms
step:1438/4740 train_time:217799ms step_avg:151.46ms
step:1439/4740 train_time:217974ms step_avg:151.48ms
step:1440/4740 train_time:218290ms step_avg:151.59ms
step:1441/4740 train_time:218465ms step_avg:151.61ms
step:1442/4740 train_time:218785ms step_avg:151.72ms
step:1443/4740 train_time:218959ms step_avg:151.74ms
step:1444/4740 train_time:219279ms step_avg:151.85ms
step:1445/4740 train_time:219456ms step_avg:151.87ms
step:1446/4740 train_time:219774ms step_avg:151.99ms
step:1447/4740 train_time:219947ms step_avg:152.00ms
step:1448/4740 train_time:220264ms step_avg:152.12ms
step:1449/4740 train_time:220441ms step_avg:152.13ms
step:1450/4740 train_time:220761ms step_avg:152.25ms
step:1451/4740 train_time:220936ms step_avg:152.26ms
step:1452/4740 train_time:221252ms step_avg:152.38ms
step:1453/4740 train_time:221426ms step_avg:152.39ms
step:1454/4740 train_time:221744ms step_avg:152.51ms
step:1455/4740 train_time:221918ms step_avg:152.52ms
step:1456/4740 train_time:222236ms step_avg:152.63ms
step:1457/4740 train_time:222412ms step_avg:152.65ms
step:1458/4740 train_time:222729ms step_avg:152.76ms
step:1459/4740 train_time:222905ms step_avg:152.78ms
step:1460/4740 train_time:223223ms step_avg:152.89ms
step:1461/4740 train_time:223398ms step_avg:152.91ms
step:1462/4740 train_time:223714ms step_avg:153.02ms
step:1463/4740 train_time:223890ms step_avg:153.03ms
step:1464/4740 train_time:224210ms step_avg:153.15ms
step:1465/4740 train_time:224387ms step_avg:153.17ms
step:1466/4740 train_time:224704ms step_avg:153.28ms
step:1467/4740 train_time:224883ms step_avg:153.29ms
step:1468/4740 train_time:225200ms step_avg:153.41ms
step:1469/4740 train_time:225377ms step_avg:153.42ms
step:1470/4740 train_time:225692ms step_avg:153.53ms
step:1471/4740 train_time:225869ms step_avg:153.55ms
step:1472/4740 train_time:226190ms step_avg:153.66ms
step:1473/4740 train_time:226365ms step_avg:153.68ms
step:1474/4740 train_time:226683ms step_avg:153.79ms
step:1475/4740 train_time:226856ms step_avg:153.80ms
step:1476/4740 train_time:227170ms step_avg:153.91ms
step:1477/4740 train_time:227346ms step_avg:153.92ms
step:1478/4740 train_time:227665ms step_avg:154.04ms
step:1479/4740 train_time:227840ms step_avg:154.05ms
step:1480/4740 train_time:228156ms step_avg:154.16ms
step:1481/4740 train_time:228333ms step_avg:154.18ms
step:1482/4740 train_time:228650ms step_avg:154.28ms
step:1483/4740 train_time:228824ms step_avg:154.30ms
step:1484/4740 train_time:229143ms step_avg:154.41ms
step:1485/4740 train_time:229318ms step_avg:154.42ms
step:1486/4740 train_time:229635ms step_avg:154.53ms
step:1487/4740 train_time:229811ms step_avg:154.55ms
step:1488/4740 train_time:230127ms step_avg:154.66ms
step:1489/4740 train_time:230304ms step_avg:154.67ms
step:1490/4740 train_time:230623ms step_avg:154.78ms
step:1491/4740 train_time:230795ms step_avg:154.79ms
step:1492/4740 train_time:231111ms step_avg:154.90ms
step:1493/4740 train_time:231287ms step_avg:154.91ms
step:1494/4740 train_time:231604ms step_avg:155.02ms
step:1495/4740 train_time:231780ms step_avg:155.04ms
step:1496/4740 train_time:232099ms step_avg:155.15ms
step:1497/4740 train_time:232273ms step_avg:155.16ms
step:1498/4740 train_time:232593ms step_avg:155.27ms
step:1499/4740 train_time:232765ms step_avg:155.28ms
step:1500/4740 train_time:233085ms step_avg:155.39ms
step:1500/4740 val_loss:3.3339 train_time:233206ms step_avg:155.47ms
step:1501/4740 train_time:233264ms step_avg:155.41ms
step:1502/4740 train_time:233592ms step_avg:155.52ms
step:1503/4740 train_time:233765ms step_avg:155.53ms
step:1504/4740 train_time:234078ms step_avg:155.64ms
step:1505/4740 train_time:234250ms step_avg:155.65ms
step:1506/4740 train_time:234573ms step_avg:155.76ms
step:1507/4740 train_time:234751ms step_avg:155.77ms
step:1508/4740 train_time:235069ms step_avg:155.88ms
step:1509/4740 train_time:235239ms step_avg:155.89ms
step:1510/4740 train_time:235560ms step_avg:156.00ms
step:1511/4740 train_time:235738ms step_avg:156.01ms
step:1512/4740 train_time:236053ms step_avg:156.12ms
step:1513/4740 train_time:236232ms step_avg:156.13ms
step:1514/4740 train_time:236546ms step_avg:156.24ms
step:1515/4740 train_time:236723ms step_avg:156.25ms
step:1516/4740 train_time:237038ms step_avg:156.36ms
step:1517/4740 train_time:237216ms step_avg:156.37ms
step:1518/4740 train_time:237533ms step_avg:156.48ms
step:1519/4740 train_time:237711ms step_avg:156.49ms
step:1520/4740 train_time:238029ms step_avg:156.60ms
step:1521/4740 train_time:238205ms step_avg:156.61ms
step:1522/4740 train_time:238520ms step_avg:156.71ms
step:1523/4740 train_time:238698ms step_avg:156.73ms
step:1524/4740 train_time:239016ms step_avg:156.83ms
step:1525/4740 train_time:239192ms step_avg:156.85ms
step:1526/4740 train_time:239509ms step_avg:156.95ms
step:1527/4740 train_time:239684ms step_avg:156.96ms
step:1528/4740 train_time:240004ms step_avg:157.07ms
step:1529/4740 train_time:240177ms step_avg:157.08ms
step:1530/4740 train_time:240495ms step_avg:157.19ms
step:1531/4740 train_time:240669ms step_avg:157.20ms
step:1532/4740 train_time:240992ms step_avg:157.31ms
step:1533/4740 train_time:241166ms step_avg:157.32ms
step:1534/4740 train_time:241481ms step_avg:157.42ms
step:1535/4740 train_time:241657ms step_avg:157.43ms
step:1536/4740 train_time:241976ms step_avg:157.54ms
step:1537/4740 train_time:242152ms step_avg:157.55ms
step:1538/4740 train_time:242469ms step_avg:157.65ms
step:1539/4740 train_time:242644ms step_avg:157.66ms
step:1540/4740 train_time:242960ms step_avg:157.77ms
step:1541/4740 train_time:243137ms step_avg:157.78ms
step:1542/4740 train_time:243456ms step_avg:157.88ms
step:1543/4740 train_time:243631ms step_avg:157.89ms
step:1544/4740 train_time:243949ms step_avg:158.00ms
step:1545/4740 train_time:244124ms step_avg:158.01ms
step:1546/4740 train_time:244441ms step_avg:158.11ms
step:1547/4740 train_time:244618ms step_avg:158.12ms
step:1548/4740 train_time:244937ms step_avg:158.23ms
step:1549/4740 train_time:245112ms step_avg:158.24ms
step:1550/4740 train_time:245425ms step_avg:158.34ms
step:1551/4740 train_time:245601ms step_avg:158.35ms
step:1552/4740 train_time:245920ms step_avg:158.45ms
step:1553/4740 train_time:246097ms step_avg:158.47ms
step:1554/4740 train_time:246416ms step_avg:158.57ms
step:1555/4740 train_time:246590ms step_avg:158.58ms
step:1556/4740 train_time:246908ms step_avg:158.68ms
step:1557/4740 train_time:247084ms step_avg:158.69ms
step:1558/4740 train_time:247399ms step_avg:158.79ms
step:1559/4740 train_time:247575ms step_avg:158.80ms
step:1560/4740 train_time:247895ms step_avg:158.91ms
step:1561/4740 train_time:248070ms step_avg:158.92ms
step:1562/4740 train_time:248389ms step_avg:159.02ms
step:1563/4740 train_time:248565ms step_avg:159.03ms
step:1564/4740 train_time:248878ms step_avg:159.13ms
step:1565/4740 train_time:249054ms step_avg:159.14ms
step:1566/4740 train_time:249374ms step_avg:159.24ms
step:1567/4740 train_time:249549ms step_avg:159.25ms
step:1568/4740 train_time:249868ms step_avg:159.35ms
step:1569/4740 train_time:250042ms step_avg:159.36ms
step:1570/4740 train_time:250359ms step_avg:159.46ms
step:1571/4740 train_time:250539ms step_avg:159.48ms
step:1572/4740 train_time:250857ms step_avg:159.58ms
step:1573/4740 train_time:251035ms step_avg:159.59ms
step:1574/4740 train_time:251355ms step_avg:159.69ms
step:1575/4740 train_time:251530ms step_avg:159.70ms
step:1576/4740 train_time:251849ms step_avg:159.80ms
step:1577/4740 train_time:252027ms step_avg:159.81ms
step:1578/4740 train_time:252344ms step_avg:159.91ms
step:1579/4740 train_time:252521ms step_avg:159.92ms
step:1580/4740 train_time:252839ms step_avg:160.02ms
step:1581/4740 train_time:253015ms step_avg:160.03ms
step:1582/4740 train_time:253335ms step_avg:160.14ms
step:1583/4740 train_time:253512ms step_avg:160.15ms
step:1584/4740 train_time:253832ms step_avg:160.25ms
step:1585/4740 train_time:254009ms step_avg:160.26ms
step:1586/4740 train_time:254326ms step_avg:160.36ms
step:1587/4740 train_time:254504ms step_avg:160.37ms
step:1588/4740 train_time:254821ms step_avg:160.47ms
step:1589/4740 train_time:254998ms step_avg:160.48ms
step:1590/4740 train_time:255316ms step_avg:160.58ms
step:1591/4740 train_time:255493ms step_avg:160.59ms
step:1592/4740 train_time:255813ms step_avg:160.69ms
step:1593/4740 train_time:255987ms step_avg:160.70ms
step:1594/4740 train_time:256307ms step_avg:160.79ms
step:1595/4740 train_time:256483ms step_avg:160.80ms
step:1596/4740 train_time:256802ms step_avg:160.90ms
step:1597/4740 train_time:256977ms step_avg:160.91ms
step:1598/4740 train_time:257297ms step_avg:161.01ms
step:1599/4740 train_time:257471ms step_avg:161.02ms
step:1600/4740 train_time:257792ms step_avg:161.12ms
step:1601/4740 train_time:257969ms step_avg:161.13ms
step:1602/4740 train_time:258289ms step_avg:161.23ms
step:1603/4740 train_time:258465ms step_avg:161.24ms
step:1604/4740 train_time:258784ms step_avg:161.34ms
step:1605/4740 train_time:258961ms step_avg:161.35ms
step:1606/4740 train_time:259278ms step_avg:161.44ms
step:1607/4740 train_time:259456ms step_avg:161.45ms
step:1608/4740 train_time:259775ms step_avg:161.55ms
step:1609/4740 train_time:259953ms step_avg:161.56ms
step:1610/4740 train_time:260273ms step_avg:161.66ms
step:1611/4740 train_time:260451ms step_avg:161.67ms
step:1612/4740 train_time:260770ms step_avg:161.77ms
step:1613/4740 train_time:260946ms step_avg:161.78ms
step:1614/4740 train_time:261262ms step_avg:161.87ms
step:1615/4740 train_time:261440ms step_avg:161.88ms
step:1616/4740 train_time:261758ms step_avg:161.98ms
step:1617/4740 train_time:261935ms step_avg:161.99ms
step:1618/4740 train_time:262254ms step_avg:162.09ms
step:1619/4740 train_time:262430ms step_avg:162.09ms
step:1620/4740 train_time:262748ms step_avg:162.19ms
step:1621/4740 train_time:262925ms step_avg:162.20ms
step:1622/4740 train_time:263243ms step_avg:162.30ms
step:1623/4740 train_time:263423ms step_avg:162.31ms
step:1624/4740 train_time:263742ms step_avg:162.40ms
step:1625/4740 train_time:263918ms step_avg:162.41ms
step:1626/4740 train_time:264237ms step_avg:162.51ms
step:1627/4740 train_time:264416ms step_avg:162.52ms
step:1628/4740 train_time:264733ms step_avg:162.61ms
step:1629/4740 train_time:264911ms step_avg:162.62ms
step:1630/4740 train_time:265230ms step_avg:162.72ms
step:1631/4740 train_time:265407ms step_avg:162.73ms
step:1632/4740 train_time:265725ms step_avg:162.82ms
step:1633/4740 train_time:265900ms step_avg:162.83ms
step:1634/4740 train_time:266219ms step_avg:162.92ms
step:1635/4740 train_time:266398ms step_avg:162.93ms
step:1636/4740 train_time:266718ms step_avg:163.03ms
step:1637/4740 train_time:266892ms step_avg:163.04ms
step:1638/4740 train_time:267213ms step_avg:163.13ms
step:1639/4740 train_time:267388ms step_avg:163.14ms
step:1640/4740 train_time:267709ms step_avg:163.24ms
step:1641/4740 train_time:267883ms step_avg:163.24ms
step:1642/4740 train_time:268201ms step_avg:163.34ms
step:1643/4740 train_time:268379ms step_avg:163.35ms
step:1644/4740 train_time:268701ms step_avg:163.44ms
step:1645/4740 train_time:268875ms step_avg:163.45ms
step:1646/4740 train_time:269196ms step_avg:163.55ms
step:1647/4740 train_time:269370ms step_avg:163.55ms
step:1648/4740 train_time:269691ms step_avg:163.65ms
step:1649/4740 train_time:269866ms step_avg:163.65ms
step:1650/4740 train_time:270182ms step_avg:163.75ms
step:1651/4740 train_time:270359ms step_avg:163.75ms
step:1652/4740 train_time:270676ms step_avg:163.85ms
step:1653/4740 train_time:270856ms step_avg:163.86ms
step:1654/4740 train_time:271175ms step_avg:163.95ms
step:1655/4740 train_time:271352ms step_avg:163.96ms
step:1656/4740 train_time:271672ms step_avg:164.05ms
step:1657/4740 train_time:271848ms step_avg:164.06ms
step:1658/4740 train_time:272164ms step_avg:164.15ms
step:1659/4740 train_time:272343ms step_avg:164.16ms
step:1660/4740 train_time:272661ms step_avg:164.25ms
step:1661/4740 train_time:272839ms step_avg:164.26ms
step:1662/4740 train_time:273157ms step_avg:164.35ms
step:1663/4740 train_time:273337ms step_avg:164.36ms
step:1664/4740 train_time:273655ms step_avg:164.46ms
step:1665/4740 train_time:273834ms step_avg:164.46ms
step:1666/4740 train_time:274155ms step_avg:164.56ms
step:1667/4740 train_time:274331ms step_avg:164.57ms
step:1668/4740 train_time:274646ms step_avg:164.66ms
step:1669/4740 train_time:274824ms step_avg:164.66ms
step:1670/4740 train_time:275143ms step_avg:164.76ms
step:1671/4740 train_time:275319ms step_avg:164.76ms
step:1672/4740 train_time:275637ms step_avg:164.85ms
step:1673/4740 train_time:275815ms step_avg:164.86ms
step:1674/4740 train_time:276135ms step_avg:164.96ms
step:1675/4740 train_time:276310ms step_avg:164.96ms
step:1676/4740 train_time:276630ms step_avg:165.05ms
step:1677/4740 train_time:276808ms step_avg:165.06ms
step:1678/4740 train_time:277125ms step_avg:165.15ms
step:1679/4740 train_time:277297ms step_avg:165.16ms
step:1680/4740 train_time:277617ms step_avg:165.25ms
step:1681/4740 train_time:277794ms step_avg:165.25ms
step:1682/4740 train_time:278116ms step_avg:165.35ms
step:1683/4740 train_time:278294ms step_avg:165.36ms
step:1684/4740 train_time:278611ms step_avg:165.45ms
step:1685/4740 train_time:278786ms step_avg:165.45ms
step:1686/4740 train_time:279107ms step_avg:165.54ms
step:1687/4740 train_time:279281ms step_avg:165.55ms
step:1688/4740 train_time:279598ms step_avg:165.64ms
step:1689/4740 train_time:279774ms step_avg:165.64ms
step:1690/4740 train_time:280095ms step_avg:165.74ms
step:1691/4740 train_time:280270ms step_avg:165.74ms
step:1692/4740 train_time:280585ms step_avg:165.83ms
step:1693/4740 train_time:280761ms step_avg:165.84ms
step:1694/4740 train_time:281081ms step_avg:165.93ms
step:1695/4740 train_time:281259ms step_avg:165.93ms
step:1696/4740 train_time:281575ms step_avg:166.02ms
step:1697/4740 train_time:281754ms step_avg:166.03ms
step:1698/4740 train_time:282074ms step_avg:166.12ms
step:1699/4740 train_time:282252ms step_avg:166.13ms
step:1700/4740 train_time:282572ms step_avg:166.22ms
step:1701/4740 train_time:282747ms step_avg:166.22ms
step:1702/4740 train_time:283064ms step_avg:166.31ms
step:1703/4740 train_time:283241ms step_avg:166.32ms
step:1704/4740 train_time:283558ms step_avg:166.41ms
step:1705/4740 train_time:283735ms step_avg:166.41ms
step:1706/4740 train_time:284054ms step_avg:166.50ms
step:1707/4740 train_time:284233ms step_avg:166.51ms
step:1708/4740 train_time:284554ms step_avg:166.60ms
step:1709/4740 train_time:284728ms step_avg:166.61ms
step:1710/4740 train_time:285048ms step_avg:166.70ms
step:1711/4740 train_time:285224ms step_avg:166.70ms
step:1712/4740 train_time:285542ms step_avg:166.79ms
step:1713/4740 train_time:285720ms step_avg:166.80ms
step:1714/4740 train_time:286037ms step_avg:166.88ms
step:1715/4740 train_time:286218ms step_avg:166.89ms
step:1716/4740 train_time:286537ms step_avg:166.98ms
step:1717/4740 train_time:286713ms step_avg:166.98ms
step:1718/4740 train_time:287033ms step_avg:167.07ms
step:1719/4740 train_time:287208ms step_avg:167.08ms
step:1720/4740 train_time:287526ms step_avg:167.17ms
step:1721/4740 train_time:287701ms step_avg:167.17ms
step:1722/4740 train_time:288018ms step_avg:167.26ms
step:1723/4740 train_time:288197ms step_avg:167.26ms
step:1724/4740 train_time:288515ms step_avg:167.35ms
step:1725/4740 train_time:288691ms step_avg:167.36ms
step:1726/4740 train_time:289014ms step_avg:167.45ms
step:1727/4740 train_time:289190ms step_avg:167.45ms
step:1728/4740 train_time:289510ms step_avg:167.54ms
step:1729/4740 train_time:289685ms step_avg:167.54ms
step:1730/4740 train_time:290002ms step_avg:167.63ms
step:1731/4740 train_time:290180ms step_avg:167.64ms
step:1732/4740 train_time:290501ms step_avg:167.73ms
step:1733/4740 train_time:290675ms step_avg:167.73ms
step:1734/4740 train_time:290997ms step_avg:167.82ms
step:1735/4740 train_time:291172ms step_avg:167.82ms
step:1736/4740 train_time:291492ms step_avg:167.91ms
step:1737/4740 train_time:291665ms step_avg:167.91ms
step:1738/4740 train_time:291982ms step_avg:168.00ms
step:1739/4740 train_time:292160ms step_avg:168.00ms
step:1740/4740 train_time:292477ms step_avg:168.09ms
step:1741/4740 train_time:292655ms step_avg:168.10ms
step:1742/4740 train_time:292973ms step_avg:168.18ms
step:1743/4740 train_time:293148ms step_avg:168.19ms
step:1744/4740 train_time:293464ms step_avg:168.27ms
step:1745/4740 train_time:293640ms step_avg:168.27ms
step:1746/4740 train_time:293959ms step_avg:168.36ms
step:1747/4740 train_time:294134ms step_avg:168.37ms
step:1748/4740 train_time:294454ms step_avg:168.45ms
step:1749/4740 train_time:294628ms step_avg:168.46ms
step:1750/4740 train_time:294946ms step_avg:168.54ms
step:1750/4740 val_loss:3.2769 train_time:295067ms step_avg:168.61ms
step:1751/4740 train_time:295124ms step_avg:168.55ms
step:1752/4740 train_time:295451ms step_avg:168.64ms
step:1753/4740 train_time:295626ms step_avg:168.64ms
step:1754/4740 train_time:295939ms step_avg:168.72ms
step:1755/4740 train_time:296111ms step_avg:168.72ms
step:1756/4740 train_time:296437ms step_avg:168.81ms
step:1757/4740 train_time:296615ms step_avg:168.82ms
step:1758/4740 train_time:296929ms step_avg:168.90ms
step:1759/4740 train_time:297105ms step_avg:168.91ms
step:1760/4740 train_time:297428ms step_avg:168.99ms
step:1761/4740 train_time:297605ms step_avg:169.00ms
step:1762/4740 train_time:297925ms step_avg:169.08ms
step:1763/4740 train_time:298100ms step_avg:169.09ms
step:1764/4740 train_time:298416ms step_avg:169.17ms
step:1765/4740 train_time:298593ms step_avg:169.17ms
step:1766/4740 train_time:298916ms step_avg:169.26ms
step:1767/4740 train_time:299089ms step_avg:169.26ms
step:1768/4740 train_time:299409ms step_avg:169.35ms
step:1769/4740 train_time:299585ms step_avg:169.35ms
step:1770/4740 train_time:299908ms step_avg:169.44ms
step:1771/4740 train_time:300082ms step_avg:169.44ms
step:1772/4740 train_time:300399ms step_avg:169.53ms
step:1773/4740 train_time:300574ms step_avg:169.53ms
step:1774/4740 train_time:300894ms step_avg:169.61ms
step:1775/4740 train_time:301070ms step_avg:169.62ms
step:1776/4740 train_time:301388ms step_avg:169.70ms
step:1777/4740 train_time:301565ms step_avg:169.70ms
step:1778/4740 train_time:301888ms step_avg:169.79ms
step:1779/4740 train_time:302062ms step_avg:169.79ms
step:1780/4740 train_time:302377ms step_avg:169.87ms
step:1781/4740 train_time:302553ms step_avg:169.88ms
step:1782/4740 train_time:302871ms step_avg:169.96ms
step:1783/4740 train_time:303047ms step_avg:169.96ms
step:1784/4740 train_time:303368ms step_avg:170.05ms
step:1785/4740 train_time:303542ms step_avg:170.05ms
step:1786/4740 train_time:303860ms step_avg:170.13ms
step:1787/4740 train_time:304037ms step_avg:170.14ms
step:1788/4740 train_time:304355ms step_avg:170.22ms
step:1789/4740 train_time:304532ms step_avg:170.22ms
step:1790/4740 train_time:304851ms step_avg:170.31ms
step:1791/4740 train_time:305030ms step_avg:170.31ms
step:1792/4740 train_time:305350ms step_avg:170.40ms
step:1793/4740 train_time:305528ms step_avg:170.40ms
step:1794/4740 train_time:305848ms step_avg:170.48ms
step:1795/4740 train_time:306023ms step_avg:170.49ms
step:1796/4740 train_time:306342ms step_avg:170.57ms
step:1797/4740 train_time:306518ms step_avg:170.57ms
step:1798/4740 train_time:306835ms step_avg:170.65ms
step:1799/4740 train_time:307014ms step_avg:170.66ms
step:1800/4740 train_time:307333ms step_avg:170.74ms
step:1801/4740 train_time:307511ms step_avg:170.74ms
step:1802/4740 train_time:307829ms step_avg:170.83ms
step:1803/4740 train_time:308005ms step_avg:170.83ms
step:1804/4740 train_time:308326ms step_avg:170.91ms
step:1805/4740 train_time:308502ms step_avg:170.92ms
step:1806/4740 train_time:308820ms step_avg:171.00ms
step:1807/4740 train_time:308997ms step_avg:171.00ms
step:1808/4740 train_time:309318ms step_avg:171.08ms
step:1809/4740 train_time:309495ms step_avg:171.09ms
step:1810/4740 train_time:309813ms step_avg:171.17ms
step:1811/4740 train_time:309991ms step_avg:171.17ms
step:1812/4740 train_time:310312ms step_avg:171.25ms
step:1813/4740 train_time:310488ms step_avg:171.26ms
step:1814/4740 train_time:310807ms step_avg:171.34ms
step:1815/4740 train_time:310983ms step_avg:171.34ms
step:1816/4740 train_time:311303ms step_avg:171.42ms
step:1817/4740 train_time:311478ms step_avg:171.42ms
step:1818/4740 train_time:311798ms step_avg:171.51ms
step:1819/4740 train_time:311972ms step_avg:171.51ms
step:1820/4740 train_time:312291ms step_avg:171.59ms
step:1821/4740 train_time:312467ms step_avg:171.59ms
step:1822/4740 train_time:312790ms step_avg:171.67ms
step:1823/4740 train_time:312967ms step_avg:171.68ms
step:1824/4740 train_time:313285ms step_avg:171.76ms
step:1825/4740 train_time:313459ms step_avg:171.76ms
step:1826/4740 train_time:313776ms step_avg:171.84ms
step:1827/4740 train_time:313956ms step_avg:171.84ms
step:1828/4740 train_time:314275ms step_avg:171.92ms
step:1829/4740 train_time:314448ms step_avg:171.92ms
step:1830/4740 train_time:314769ms step_avg:172.00ms
step:1831/4740 train_time:314943ms step_avg:172.01ms
step:1832/4740 train_time:315263ms step_avg:172.09ms
step:1833/4740 train_time:315439ms step_avg:172.09ms
step:1834/4740 train_time:315758ms step_avg:172.17ms
step:1835/4740 train_time:315933ms step_avg:172.17ms
step:1836/4740 train_time:316251ms step_avg:172.25ms
step:1837/4740 train_time:316429ms step_avg:172.25ms
step:1838/4740 train_time:316748ms step_avg:172.33ms
step:1839/4740 train_time:316927ms step_avg:172.34ms
step:1840/4740 train_time:317247ms step_avg:172.42ms
step:1841/4740 train_time:317423ms step_avg:172.42ms
step:1842/4740 train_time:317740ms step_avg:172.50ms
step:1843/4740 train_time:317920ms step_avg:172.50ms
step:1844/4740 train_time:318236ms step_avg:172.58ms
step:1845/4740 train_time:318412ms step_avg:172.58ms
step:1846/4740 train_time:318728ms step_avg:172.66ms
step:1847/4740 train_time:318908ms step_avg:172.66ms
step:1848/4740 train_time:319227ms step_avg:172.74ms
step:1849/4740 train_time:319402ms step_avg:172.74ms
step:1850/4740 train_time:319720ms step_avg:172.82ms
step:1851/4740 train_time:319898ms step_avg:172.82ms
step:1852/4740 train_time:320216ms step_avg:172.90ms
step:1853/4740 train_time:320391ms step_avg:172.90ms
step:1854/4740 train_time:320709ms step_avg:172.98ms
step:1855/4740 train_time:320886ms step_avg:172.98ms
step:1856/4740 train_time:321207ms step_avg:173.06ms
step:1857/4740 train_time:321383ms step_avg:173.07ms
step:1858/4740 train_time:321703ms step_avg:173.15ms
step:1859/4740 train_time:321880ms step_avg:173.15ms
step:1860/4740 train_time:322198ms step_avg:173.22ms
step:1861/4740 train_time:322373ms step_avg:173.23ms
step:1862/4740 train_time:322694ms step_avg:173.30ms
step:1863/4740 train_time:322869ms step_avg:173.31ms
step:1864/4740 train_time:323190ms step_avg:173.38ms
step:1865/4740 train_time:323368ms step_avg:173.39ms
step:1866/4740 train_time:323686ms step_avg:173.47ms
step:1867/4740 train_time:323861ms step_avg:173.47ms
step:1868/4740 train_time:324176ms step_avg:173.54ms
step:1869/4740 train_time:324355ms step_avg:173.54ms
step:1870/4740 train_time:324672ms step_avg:173.62ms
step:1871/4740 train_time:324849ms step_avg:173.62ms
step:1872/4740 train_time:325169ms step_avg:173.70ms
step:1873/4740 train_time:325345ms step_avg:173.70ms
step:1874/4740 train_time:325666ms step_avg:173.78ms
step:1875/4740 train_time:325840ms step_avg:173.78ms
step:1876/4740 train_time:326159ms step_avg:173.86ms
step:1877/4740 train_time:326333ms step_avg:173.86ms
step:1878/4740 train_time:326655ms step_avg:173.94ms
step:1879/4740 train_time:326831ms step_avg:173.94ms
step:1880/4740 train_time:327149ms step_avg:174.02ms
step:1881/4740 train_time:327325ms step_avg:174.02ms
step:1882/4740 train_time:327643ms step_avg:174.09ms
step:1883/4740 train_time:327821ms step_avg:174.10ms
step:1884/4740 train_time:328136ms step_avg:174.17ms
step:1885/4740 train_time:328315ms step_avg:174.17ms
step:1886/4740 train_time:328632ms step_avg:174.25ms
step:1887/4740 train_time:328810ms step_avg:174.25ms
step:1888/4740 train_time:329130ms step_avg:174.33ms
step:1889/4740 train_time:329305ms step_avg:174.33ms
step:1890/4740 train_time:329625ms step_avg:174.40ms
step:1891/4740 train_time:329800ms step_avg:174.40ms
step:1892/4740 train_time:330117ms step_avg:174.48ms
step:1893/4740 train_time:330295ms step_avg:174.48ms
step:1894/4740 train_time:330616ms step_avg:174.56ms
step:1895/4740 train_time:330793ms step_avg:174.56ms
step:1896/4740 train_time:331113ms step_avg:174.64ms
step:1897/4740 train_time:331289ms step_avg:174.64ms
step:1898/4740 train_time:331608ms step_avg:174.71ms
step:1899/4740 train_time:331783ms step_avg:174.71ms
step:1900/4740 train_time:332105ms step_avg:174.79ms
step:1901/4740 train_time:332280ms step_avg:174.79ms
step:1902/4740 train_time:332598ms step_avg:174.87ms
step:1903/4740 train_time:332773ms step_avg:174.87ms
step:1904/4740 train_time:333095ms step_avg:174.94ms
step:1905/4740 train_time:333270ms step_avg:174.94ms
step:1906/4740 train_time:333589ms step_avg:175.02ms
step:1907/4740 train_time:333767ms step_avg:175.02ms
step:1908/4740 train_time:334088ms step_avg:175.10ms
step:1909/4740 train_time:334263ms step_avg:175.10ms
step:1910/4740 train_time:334584ms step_avg:175.17ms
step:1911/4740 train_time:334763ms step_avg:175.18ms
step:1912/4740 train_time:335082ms step_avg:175.25ms
step:1913/4740 train_time:335258ms step_avg:175.25ms
step:1914/4740 train_time:335575ms step_avg:175.33ms
step:1915/4740 train_time:335751ms step_avg:175.33ms
step:1916/4740 train_time:336070ms step_avg:175.40ms
step:1917/4740 train_time:336247ms step_avg:175.40ms
step:1918/4740 train_time:336567ms step_avg:175.48ms
step:1919/4740 train_time:336743ms step_avg:175.48ms
step:1920/4740 train_time:337060ms step_avg:175.55ms
step:1921/4740 train_time:337237ms step_avg:175.55ms
step:1922/4740 train_time:337555ms step_avg:175.63ms
step:1923/4740 train_time:337733ms step_avg:175.63ms
step:1924/4740 train_time:338052ms step_avg:175.70ms
step:1925/4740 train_time:338229ms step_avg:175.70ms
step:1926/4740 train_time:338548ms step_avg:175.78ms
step:1927/4740 train_time:338722ms step_avg:175.78ms
step:1928/4740 train_time:339039ms step_avg:175.85ms
step:1929/4740 train_time:339218ms step_avg:175.85ms
step:1930/4740 train_time:339534ms step_avg:175.92ms
step:1931/4740 train_time:339713ms step_avg:175.93ms
step:1932/4740 train_time:340033ms step_avg:176.00ms
step:1933/4740 train_time:340210ms step_avg:176.00ms
step:1934/4740 train_time:340530ms step_avg:176.08ms
step:1935/4740 train_time:340708ms step_avg:176.08ms
step:1936/4740 train_time:341028ms step_avg:176.15ms
step:1937/4740 train_time:341204ms step_avg:176.15ms
step:1938/4740 train_time:341524ms step_avg:176.23ms
step:1939/4740 train_time:341700ms step_avg:176.22ms
step:1940/4740 train_time:342015ms step_avg:176.30ms
step:1941/4740 train_time:342190ms step_avg:176.30ms
step:1942/4740 train_time:342510ms step_avg:176.37ms
step:1943/4740 train_time:342685ms step_avg:176.37ms
step:1944/4740 train_time:343005ms step_avg:176.44ms
step:1945/4740 train_time:343179ms step_avg:176.44ms
step:1946/4740 train_time:343497ms step_avg:176.51ms
step:1947/4740 train_time:343674ms step_avg:176.51ms
step:1948/4740 train_time:343995ms step_avg:176.59ms
step:1949/4740 train_time:344170ms step_avg:176.59ms
step:1950/4740 train_time:344490ms step_avg:176.66ms
step:1951/4740 train_time:344666ms step_avg:176.66ms
step:1952/4740 train_time:344988ms step_avg:176.74ms
step:1953/4740 train_time:345162ms step_avg:176.73ms
step:1954/4740 train_time:345485ms step_avg:176.81ms
step:1955/4740 train_time:345662ms step_avg:176.81ms
step:1956/4740 train_time:345982ms step_avg:176.88ms
step:1957/4740 train_time:346158ms step_avg:176.88ms
step:1958/4740 train_time:346475ms step_avg:176.95ms
step:1959/4740 train_time:346651ms step_avg:176.95ms
step:1960/4740 train_time:346969ms step_avg:177.03ms
step:1961/4740 train_time:347147ms step_avg:177.03ms
step:1962/4740 train_time:347469ms step_avg:177.10ms
step:1963/4740 train_time:347645ms step_avg:177.10ms
step:1964/4740 train_time:347965ms step_avg:177.17ms
step:1965/4740 train_time:348144ms step_avg:177.17ms
step:1966/4740 train_time:348463ms step_avg:177.24ms
step:1967/4740 train_time:348636ms step_avg:177.24ms
step:1968/4740 train_time:348957ms step_avg:177.32ms
step:1969/4740 train_time:349134ms step_avg:177.32ms
step:1970/4740 train_time:349453ms step_avg:177.39ms
step:1971/4740 train_time:349628ms step_avg:177.39ms
step:1972/4740 train_time:349949ms step_avg:177.46ms
step:1973/4740 train_time:350126ms step_avg:177.46ms
step:1974/4740 train_time:350443ms step_avg:177.53ms
step:1975/4740 train_time:350622ms step_avg:177.53ms
step:1976/4740 train_time:350943ms step_avg:177.60ms
step:1977/4740 train_time:351118ms step_avg:177.60ms
step:1978/4740 train_time:351436ms step_avg:177.67ms
step:1979/4740 train_time:351612ms step_avg:177.67ms
step:1980/4740 train_time:351932ms step_avg:177.74ms
step:1981/4740 train_time:352109ms step_avg:177.74ms
step:1982/4740 train_time:352430ms step_avg:177.82ms
step:1983/4740 train_time:352607ms step_avg:177.81ms
step:1984/4740 train_time:352926ms step_avg:177.89ms
step:1985/4740 train_time:353104ms step_avg:177.89ms
step:1986/4740 train_time:353423ms step_avg:177.96ms
step:1987/4740 train_time:353602ms step_avg:177.96ms
step:1988/4740 train_time:353919ms step_avg:178.03ms
step:1989/4740 train_time:354097ms step_avg:178.03ms
step:1990/4740 train_time:354415ms step_avg:178.10ms
step:1991/4740 train_time:354593ms step_avg:178.10ms
step:1992/4740 train_time:354913ms step_avg:178.17ms
step:1993/4740 train_time:355088ms step_avg:178.17ms
step:1994/4740 train_time:355409ms step_avg:178.24ms
step:1995/4740 train_time:355584ms step_avg:178.24ms
step:1996/4740 train_time:355907ms step_avg:178.31ms
step:1997/4740 train_time:356087ms step_avg:178.31ms
step:1998/4740 train_time:356410ms step_avg:178.38ms
step:1999/4740 train_time:356585ms step_avg:178.38ms
step:2000/4740 train_time:356906ms step_avg:178.45ms
step:2000/4740 val_loss:3.2324 train_time:357032ms step_avg:178.52ms
step:2001/4740 train_time:357089ms step_avg:178.46ms
step:2002/4740 train_time:357423ms step_avg:178.53ms
step:2003/4740 train_time:357599ms step_avg:178.53ms
step:2004/4740 train_time:357911ms step_avg:178.60ms
step:2005/4740 train_time:358085ms step_avg:178.60ms
step:2006/4740 train_time:358412ms step_avg:178.67ms
step:2007/4740 train_time:358591ms step_avg:178.67ms
step:2008/4740 train_time:358906ms step_avg:178.74ms
step:2009/4740 train_time:359082ms step_avg:178.74ms
step:2010/4740 train_time:359407ms step_avg:178.81ms
step:2011/4740 train_time:359584ms step_avg:178.81ms
step:2012/4740 train_time:359901ms step_avg:178.88ms
step:2013/4740 train_time:360075ms step_avg:178.88ms
step:2014/4740 train_time:360392ms step_avg:178.94ms
step:2015/4740 train_time:360573ms step_avg:178.94ms
step:2016/4740 train_time:360892ms step_avg:179.01ms
step:2017/4740 train_time:361068ms step_avg:179.01ms
step:2018/4740 train_time:361388ms step_avg:179.08ms
step:2019/4740 train_time:361564ms step_avg:179.08ms
step:2020/4740 train_time:361884ms step_avg:179.15ms
step:2021/4740 train_time:362061ms step_avg:179.15ms
step:2022/4740 train_time:362382ms step_avg:179.22ms
step:2023/4740 train_time:362558ms step_avg:179.22ms
step:2024/4740 train_time:362877ms step_avg:179.29ms
step:2025/4740 train_time:363055ms step_avg:179.29ms
step:2026/4740 train_time:363372ms step_avg:179.35ms
step:2027/4740 train_time:363552ms step_avg:179.35ms
step:2028/4740 train_time:363871ms step_avg:179.42ms
step:2029/4740 train_time:364050ms step_avg:179.42ms
step:2030/4740 train_time:364367ms step_avg:179.49ms
step:2031/4740 train_time:364546ms step_avg:179.49ms
step:2032/4740 train_time:364866ms step_avg:179.56ms
step:2033/4740 train_time:365043ms step_avg:179.56ms
step:2034/4740 train_time:365365ms step_avg:179.63ms
step:2035/4740 train_time:365543ms step_avg:179.63ms
step:2036/4740 train_time:365861ms step_avg:179.70ms
step:2037/4740 train_time:366037ms step_avg:179.69ms
step:2038/4740 train_time:366357ms step_avg:179.76ms
step:2039/4740 train_time:366532ms step_avg:179.76ms
step:2040/4740 train_time:366849ms step_avg:179.83ms
step:2041/4740 train_time:367028ms step_avg:179.83ms
step:2042/4740 train_time:367347ms step_avg:179.90ms
step:2043/4740 train_time:367524ms step_avg:179.89ms
step:2044/4740 train_time:367844ms step_avg:179.96ms
step:2045/4740 train_time:368018ms step_avg:179.96ms
step:2046/4740 train_time:368338ms step_avg:180.03ms
step:2047/4740 train_time:368514ms step_avg:180.03ms
step:2048/4740 train_time:368830ms step_avg:180.09ms
step:2049/4740 train_time:369009ms step_avg:180.09ms
step:2050/4740 train_time:369329ms step_avg:180.16ms
step:2051/4740 train_time:369507ms step_avg:180.16ms
step:2052/4740 train_time:369828ms step_avg:180.23ms
step:2053/4740 train_time:370005ms step_avg:180.23ms
step:2054/4740 train_time:370326ms step_avg:180.29ms
step:2055/4740 train_time:370503ms step_avg:180.29ms
step:2056/4740 train_time:370824ms step_avg:180.36ms
step:2057/4740 train_time:370999ms step_avg:180.36ms
step:2058/4740 train_time:371317ms step_avg:180.43ms
step:2059/4740 train_time:371500ms step_avg:180.43ms
step:2060/4740 train_time:371819ms step_avg:180.49ms
step:2061/4740 train_time:371997ms step_avg:180.49ms
step:2062/4740 train_time:372314ms step_avg:180.56ms
step:2063/4740 train_time:372491ms step_avg:180.56ms
step:2064/4740 train_time:372809ms step_avg:180.62ms
step:2065/4740 train_time:372986ms step_avg:180.62ms
step:2066/4740 train_time:373309ms step_avg:180.69ms
step:2067/4740 train_time:373486ms step_avg:180.69ms
step:2068/4740 train_time:373806ms step_avg:180.76ms
step:2069/4740 train_time:373981ms step_avg:180.75ms
step:2070/4740 train_time:374300ms step_avg:180.82ms
step:2071/4740 train_time:374479ms step_avg:180.82ms
step:2072/4740 train_time:374798ms step_avg:180.89ms
step:2073/4740 train_time:374979ms step_avg:180.89ms
step:2074/4740 train_time:375300ms step_avg:180.95ms
step:2075/4740 train_time:375476ms step_avg:180.95ms
step:2076/4740 train_time:375794ms step_avg:181.02ms
step:2077/4740 train_time:375971ms step_avg:181.02ms
step:2078/4740 train_time:376290ms step_avg:181.08ms
step:2079/4740 train_time:376466ms step_avg:181.08ms
step:2080/4740 train_time:376787ms step_avg:181.15ms
step:2081/4740 train_time:376966ms step_avg:181.15ms
step:2082/4740 train_time:377286ms step_avg:181.21ms
step:2083/4740 train_time:377462ms step_avg:181.21ms
step:2084/4740 train_time:377783ms step_avg:181.28ms
step:2085/4740 train_time:377961ms step_avg:181.28ms
step:2086/4740 train_time:378281ms step_avg:181.34ms
step:2087/4740 train_time:378459ms step_avg:181.34ms
step:2088/4740 train_time:378775ms step_avg:181.41ms
step:2089/4740 train_time:378955ms step_avg:181.40ms
step:2090/4740 train_time:379272ms step_avg:181.47ms
step:2091/4740 train_time:379451ms step_avg:181.47ms
step:2092/4740 train_time:379770ms step_avg:181.53ms
step:2093/4740 train_time:379949ms step_avg:181.53ms
step:2094/4740 train_time:380268ms step_avg:181.60ms
step:2095/4740 train_time:380448ms step_avg:181.60ms
step:2096/4740 train_time:380767ms step_avg:181.66ms
step:2097/4740 train_time:380945ms step_avg:181.66ms
step:2098/4740 train_time:381265ms step_avg:181.73ms
step:2099/4740 train_time:381445ms step_avg:181.73ms
step:2100/4740 train_time:381765ms step_avg:181.79ms
step:2101/4740 train_time:381946ms step_avg:181.79ms
step:2102/4740 train_time:382265ms step_avg:181.86ms
step:2103/4740 train_time:382443ms step_avg:181.86ms
step:2104/4740 train_time:382764ms step_avg:181.92ms
step:2105/4740 train_time:382942ms step_avg:181.92ms
step:2106/4740 train_time:383260ms step_avg:181.98ms
step:2107/4740 train_time:383437ms step_avg:181.98ms
step:2108/4740 train_time:383758ms step_avg:182.05ms
step:2109/4740 train_time:383935ms step_avg:182.05ms
step:2110/4740 train_time:384254ms step_avg:182.11ms
step:2111/4740 train_time:384431ms step_avg:182.11ms
step:2112/4740 train_time:384749ms step_avg:182.17ms
step:2113/4740 train_time:384930ms step_avg:182.17ms
step:2114/4740 train_time:385246ms step_avg:182.24ms
step:2115/4740 train_time:385429ms step_avg:182.24ms
step:2116/4740 train_time:385746ms step_avg:182.30ms
step:2117/4740 train_time:385924ms step_avg:182.30ms
step:2118/4740 train_time:386242ms step_avg:182.36ms
step:2119/4740 train_time:386422ms step_avg:182.36ms
step:2120/4740 train_time:386741ms step_avg:182.43ms
step:2121/4740 train_time:386919ms step_avg:182.42ms
step:2122/4740 train_time:387239ms step_avg:182.49ms
step:2123/4740 train_time:387417ms step_avg:182.49ms
step:2124/4740 train_time:387738ms step_avg:182.55ms
step:2125/4740 train_time:387914ms step_avg:182.55ms
step:2126/4740 train_time:388231ms step_avg:182.61ms
step:2127/4740 train_time:388408ms step_avg:182.61ms
step:2128/4740 train_time:388728ms step_avg:182.67ms
step:2129/4740 train_time:388906ms step_avg:182.67ms
step:2130/4740 train_time:389226ms step_avg:182.74ms
step:2131/4740 train_time:389402ms step_avg:182.73ms
step:2132/4740 train_time:389720ms step_avg:182.80ms
step:2133/4740 train_time:389900ms step_avg:182.79ms
step:2134/4740 train_time:390220ms step_avg:182.86ms
step:2135/4740 train_time:390396ms step_avg:182.86ms
step:2136/4740 train_time:390716ms step_avg:182.92ms
step:2137/4740 train_time:390894ms step_avg:182.92ms
step:2138/4740 train_time:391214ms step_avg:182.98ms
step:2139/4740 train_time:391392ms step_avg:182.98ms
step:2140/4740 train_time:391713ms step_avg:183.04ms
step:2141/4740 train_time:391888ms step_avg:183.04ms
step:2142/4740 train_time:392210ms step_avg:183.10ms
step:2143/4740 train_time:392385ms step_avg:183.10ms
step:2144/4740 train_time:392707ms step_avg:183.17ms
step:2145/4740 train_time:392883ms step_avg:183.16ms
step:2146/4740 train_time:393204ms step_avg:183.23ms
step:2147/4740 train_time:393379ms step_avg:183.22ms
step:2148/4740 train_time:393698ms step_avg:183.29ms
step:2149/4740 train_time:393873ms step_avg:183.28ms
step:2150/4740 train_time:394193ms step_avg:183.35ms
step:2151/4740 train_time:394369ms step_avg:183.34ms
step:2152/4740 train_time:394688ms step_avg:183.41ms
step:2153/4740 train_time:394863ms step_avg:183.40ms
step:2154/4740 train_time:395185ms step_avg:183.47ms
step:2155/4740 train_time:395364ms step_avg:183.46ms
step:2156/4740 train_time:395683ms step_avg:183.53ms
step:2157/4740 train_time:395861ms step_avg:183.52ms
step:2158/4740 train_time:396178ms step_avg:183.59ms
step:2159/4740 train_time:396356ms step_avg:183.58ms
step:2160/4740 train_time:396673ms step_avg:183.64ms
step:2161/4740 train_time:396854ms step_avg:183.64ms
step:2162/4740 train_time:397172ms step_avg:183.71ms
step:2163/4740 train_time:397348ms step_avg:183.70ms
step:2164/4740 train_time:397667ms step_avg:183.76ms
step:2165/4740 train_time:397845ms step_avg:183.76ms
step:2166/4740 train_time:398165ms step_avg:183.83ms
step:2167/4740 train_time:398344ms step_avg:183.82ms
step:2168/4740 train_time:398664ms step_avg:183.89ms
step:2169/4740 train_time:398844ms step_avg:183.88ms
step:2170/4740 train_time:399165ms step_avg:183.95ms
step:2171/4740 train_time:399340ms step_avg:183.94ms
step:2172/4740 train_time:399657ms step_avg:184.00ms
step:2173/4740 train_time:399836ms step_avg:184.00ms
step:2174/4740 train_time:400153ms step_avg:184.06ms
step:2175/4740 train_time:400335ms step_avg:184.06ms
step:2176/4740 train_time:400653ms step_avg:184.12ms
step:2177/4740 train_time:400831ms step_avg:184.12ms
step:2178/4740 train_time:401150ms step_avg:184.18ms
step:2179/4740 train_time:401326ms step_avg:184.18ms
step:2180/4740 train_time:401647ms step_avg:184.24ms
step:2181/4740 train_time:401822ms step_avg:184.24ms
step:2182/4740 train_time:402142ms step_avg:184.30ms
step:2183/4740 train_time:402318ms step_avg:184.30ms
step:2184/4740 train_time:402636ms step_avg:184.36ms
step:2185/4740 train_time:402814ms step_avg:184.35ms
step:2186/4740 train_time:403133ms step_avg:184.42ms
step:2187/4740 train_time:403311ms step_avg:184.41ms
step:2188/4740 train_time:403628ms step_avg:184.47ms
step:2189/4740 train_time:403806ms step_avg:184.47ms
step:2190/4740 train_time:404125ms step_avg:184.53ms
step:2191/4740 train_time:404302ms step_avg:184.53ms
step:2192/4740 train_time:404620ms step_avg:184.59ms
step:2193/4740 train_time:404799ms step_avg:184.59ms
step:2194/4740 train_time:405115ms step_avg:184.65ms
step:2195/4740 train_time:405294ms step_avg:184.64ms
step:2196/4740 train_time:405616ms step_avg:184.71ms
step:2197/4740 train_time:405791ms step_avg:184.70ms
step:2198/4740 train_time:406111ms step_avg:184.76ms
step:2199/4740 train_time:406289ms step_avg:184.76ms
step:2200/4740 train_time:406607ms step_avg:184.82ms
step:2201/4740 train_time:406783ms step_avg:184.82ms
step:2202/4740 train_time:407105ms step_avg:184.88ms
step:2203/4740 train_time:407282ms step_avg:184.88ms
step:2204/4740 train_time:407600ms step_avg:184.94ms
step:2205/4740 train_time:407778ms step_avg:184.93ms
step:2206/4740 train_time:408100ms step_avg:185.00ms
step:2207/4740 train_time:408276ms step_avg:184.99ms
step:2208/4740 train_time:408594ms step_avg:185.05ms
step:2209/4740 train_time:408769ms step_avg:185.05ms
step:2210/4740 train_time:409088ms step_avg:185.11ms
step:2211/4740 train_time:409265ms step_avg:185.10ms
step:2212/4740 train_time:409588ms step_avg:185.17ms
step:2213/4740 train_time:409762ms step_avg:185.16ms
step:2214/4740 train_time:410078ms step_avg:185.22ms
step:2215/4740 train_time:410256ms step_avg:185.22ms
step:2216/4740 train_time:410572ms step_avg:185.28ms
step:2217/4740 train_time:410756ms step_avg:185.28ms
step:2218/4740 train_time:411073ms step_avg:185.33ms
step:2219/4740 train_time:411250ms step_avg:185.33ms
step:2220/4740 train_time:411572ms step_avg:185.39ms
step:2221/4740 train_time:411749ms step_avg:185.39ms
step:2222/4740 train_time:412070ms step_avg:185.45ms
step:2223/4740 train_time:412249ms step_avg:185.45ms
step:2224/4740 train_time:412570ms step_avg:185.51ms
step:2225/4740 train_time:412745ms step_avg:185.50ms
step:2226/4740 train_time:413066ms step_avg:185.56ms
step:2227/4740 train_time:413241ms step_avg:185.56ms
step:2228/4740 train_time:413563ms step_avg:185.62ms
step:2229/4740 train_time:413740ms step_avg:185.62ms
step:2230/4740 train_time:414061ms step_avg:185.68ms
step:2231/4740 train_time:414238ms step_avg:185.67ms
step:2232/4740 train_time:414559ms step_avg:185.73ms
step:2233/4740 train_time:414733ms step_avg:185.73ms
step:2234/4740 train_time:415050ms step_avg:185.79ms
step:2235/4740 train_time:415228ms step_avg:185.78ms
step:2236/4740 train_time:415549ms step_avg:185.84ms
step:2237/4740 train_time:415725ms step_avg:185.84ms
step:2238/4740 train_time:416041ms step_avg:185.90ms
step:2239/4740 train_time:416224ms step_avg:185.90ms
step:2240/4740 train_time:416546ms step_avg:185.96ms
step:2241/4740 train_time:416722ms step_avg:185.95ms
step:2242/4740 train_time:417044ms step_avg:186.01ms
step:2243/4740 train_time:417221ms step_avg:186.01ms
step:2244/4740 train_time:417541ms step_avg:186.07ms
step:2245/4740 train_time:417717ms step_avg:186.07ms
step:2246/4740 train_time:418037ms step_avg:186.12ms
step:2247/4740 train_time:418213ms step_avg:186.12ms
step:2248/4740 train_time:418531ms step_avg:186.18ms
step:2249/4740 train_time:418706ms step_avg:186.17ms
step:2250/4740 train_time:419026ms step_avg:186.23ms
step:2250/4740 val_loss:3.1953 train_time:419149ms step_avg:186.29ms
step:2251/4740 train_time:419205ms step_avg:186.23ms
step:2252/4740 train_time:419537ms step_avg:186.30ms
step:2253/4740 train_time:419713ms step_avg:186.29ms
step:2254/4740 train_time:420028ms step_avg:186.35ms
step:2255/4740 train_time:420202ms step_avg:186.34ms
step:2256/4740 train_time:420527ms step_avg:186.40ms
step:2257/4740 train_time:420704ms step_avg:186.40ms
step:2258/4740 train_time:421023ms step_avg:186.46ms
step:2259/4740 train_time:421199ms step_avg:186.45ms
step:2260/4740 train_time:421520ms step_avg:186.51ms
step:2261/4740 train_time:421699ms step_avg:186.51ms
step:2262/4740 train_time:422019ms step_avg:186.57ms
step:2263/4740 train_time:422196ms step_avg:186.56ms
step:2264/4740 train_time:422514ms step_avg:186.62ms
step:2265/4740 train_time:422690ms step_avg:186.62ms
step:2266/4740 train_time:423011ms step_avg:186.68ms
step:2267/4740 train_time:423186ms step_avg:186.67ms
step:2268/4740 train_time:423505ms step_avg:186.73ms
step:2269/4740 train_time:423681ms step_avg:186.73ms
step:2270/4740 train_time:424003ms step_avg:186.79ms
step:2271/4740 train_time:424179ms step_avg:186.78ms
step:2272/4740 train_time:424501ms step_avg:186.84ms
step:2273/4740 train_time:424676ms step_avg:186.84ms
step:2274/4740 train_time:424995ms step_avg:186.89ms
step:2275/4740 train_time:425169ms step_avg:186.89ms
step:2276/4740 train_time:425491ms step_avg:186.95ms
step:2277/4740 train_time:425667ms step_avg:186.94ms
step:2278/4740 train_time:425985ms step_avg:187.00ms
step:2279/4740 train_time:426163ms step_avg:187.00ms
step:2280/4740 train_time:426483ms step_avg:187.05ms
step:2281/4740 train_time:426662ms step_avg:187.05ms
step:2282/4740 train_time:426984ms step_avg:187.11ms
step:2283/4740 train_time:427159ms step_avg:187.10ms
step:2284/4740 train_time:427478ms step_avg:187.16ms
step:2285/4740 train_time:427655ms step_avg:187.16ms
step:2286/4740 train_time:427975ms step_avg:187.22ms
step:2287/4740 train_time:428151ms step_avg:187.21ms
step:2288/4740 train_time:428468ms step_avg:187.27ms
step:2289/4740 train_time:428644ms step_avg:187.26ms
step:2290/4740 train_time:428964ms step_avg:187.32ms
step:2291/4740 train_time:429144ms step_avg:187.32ms
step:2292/4740 train_time:429464ms step_avg:187.38ms
step:2293/4740 train_time:429641ms step_avg:187.37ms
step:2294/4740 train_time:429963ms step_avg:187.43ms
step:2295/4740 train_time:430137ms step_avg:187.42ms
step:2296/4740 train_time:430454ms step_avg:187.48ms
step:2297/4740 train_time:430630ms step_avg:187.47ms
step:2298/4740 train_time:430947ms step_avg:187.53ms
step:2299/4740 train_time:431126ms step_avg:187.53ms
step:2300/4740 train_time:431445ms step_avg:187.58ms
step:2301/4740 train_time:431623ms step_avg:187.58ms
step:2302/4740 train_time:431944ms step_avg:187.64ms
step:2303/4740 train_time:432122ms step_avg:187.63ms
step:2304/4740 train_time:432443ms step_avg:187.69ms
step:2305/4740 train_time:432620ms step_avg:187.69ms
step:2306/4740 train_time:432938ms step_avg:187.74ms
step:2307/4740 train_time:433117ms step_avg:187.74ms
step:2308/4740 train_time:433437ms step_avg:187.80ms
step:2309/4740 train_time:433612ms step_avg:187.79ms
step:2310/4740 train_time:433931ms step_avg:187.85ms
step:2311/4740 train_time:434107ms step_avg:187.84ms
step:2312/4740 train_time:434426ms step_avg:187.90ms
step:2313/4740 train_time:434607ms step_avg:187.90ms
step:2314/4740 train_time:434924ms step_avg:187.95ms
step:2315/4740 train_time:435101ms step_avg:187.95ms
step:2316/4740 train_time:435421ms step_avg:188.01ms
step:2317/4740 train_time:435598ms step_avg:188.00ms
step:2318/4740 train_time:435920ms step_avg:188.06ms
step:2319/4740 train_time:436096ms step_avg:188.05ms
step:2320/4740 train_time:436416ms step_avg:188.11ms
step:2321/4740 train_time:436594ms step_avg:188.11ms
step:2322/4740 train_time:436913ms step_avg:188.16ms
step:2323/4740 train_time:437088ms step_avg:188.16ms
step:2324/4740 train_time:437412ms step_avg:188.21ms
step:2325/4740 train_time:437589ms step_avg:188.21ms
step:2326/4740 train_time:437908ms step_avg:188.27ms
step:2327/4740 train_time:438085ms step_avg:188.26ms
step:2328/4740 train_time:438405ms step_avg:188.32ms
step:2329/4740 train_time:438581ms step_avg:188.31ms
step:2330/4740 train_time:438901ms step_avg:188.37ms
step:2331/4740 train_time:439077ms step_avg:188.36ms
step:2332/4740 train_time:439395ms step_avg:188.42ms
step:2333/4740 train_time:439571ms step_avg:188.41ms
step:2334/4740 train_time:439891ms step_avg:188.47ms
step:2335/4740 train_time:440068ms step_avg:188.47ms
step:2336/4740 train_time:440387ms step_avg:188.52ms
step:2337/4740 train_time:440564ms step_avg:188.52ms
step:2338/4740 train_time:440887ms step_avg:188.57ms
step:2339/4740 train_time:441062ms step_avg:188.57ms
step:2340/4740 train_time:441386ms step_avg:188.63ms
step:2341/4740 train_time:441562ms step_avg:188.62ms
step:2342/4740 train_time:441884ms step_avg:188.68ms
step:2343/4740 train_time:442060ms step_avg:188.67ms
step:2344/4740 train_time:442381ms step_avg:188.73ms
step:2345/4740 train_time:442556ms step_avg:188.72ms
step:2346/4740 train_time:442875ms step_avg:188.78ms
step:2347/4740 train_time:443052ms step_avg:188.77ms
step:2348/4740 train_time:443368ms step_avg:188.83ms
step:2349/4740 train_time:443545ms step_avg:188.82ms
step:2350/4740 train_time:443863ms step_avg:188.88ms
step:2351/4740 train_time:444041ms step_avg:188.87ms
step:2352/4740 train_time:444362ms step_avg:188.93ms
step:2353/4740 train_time:444540ms step_avg:188.92ms
step:2354/4740 train_time:444860ms step_avg:188.98ms
step:2355/4740 train_time:445041ms step_avg:188.98ms
step:2356/4740 train_time:445361ms step_avg:189.03ms
step:2357/4740 train_time:445535ms step_avg:189.03ms
step:2358/4740 train_time:445854ms step_avg:189.08ms
step:2359/4740 train_time:446037ms step_avg:189.08ms
step:2360/4740 train_time:446352ms step_avg:189.13ms
step:2361/4740 train_time:446530ms step_avg:189.13ms
step:2362/4740 train_time:446850ms step_avg:189.18ms
step:2363/4740 train_time:447030ms step_avg:189.18ms
step:2364/4740 train_time:447349ms step_avg:189.23ms
step:2365/4740 train_time:447525ms step_avg:189.23ms
step:2366/4740 train_time:447847ms step_avg:189.28ms
step:2367/4740 train_time:448026ms step_avg:189.28ms
step:2368/4740 train_time:448345ms step_avg:189.33ms
step:2369/4740 train_time:448521ms step_avg:189.33ms
step:2370/4740 train_time:448844ms step_avg:189.39ms
step:2371/4740 train_time:449024ms step_avg:189.38ms
step:2372/4740 train_time:449344ms step_avg:189.44ms
step:2373/4740 train_time:449520ms step_avg:189.43ms
step:2374/4740 train_time:449844ms step_avg:189.49ms
step:2375/4740 train_time:450020ms step_avg:189.48ms
step:2376/4740 train_time:450341ms step_avg:189.54ms
step:2377/4740 train_time:450516ms step_avg:189.53ms
step:2378/4740 train_time:450833ms step_avg:189.59ms
step:2379/4740 train_time:451012ms step_avg:189.58ms
step:2380/4740 train_time:451331ms step_avg:189.63ms
step:2381/4740 train_time:451509ms step_avg:189.63ms
step:2382/4740 train_time:451829ms step_avg:189.68ms
step:2383/4740 train_time:452007ms step_avg:189.68ms
step:2384/4740 train_time:452327ms step_avg:189.73ms
step:2385/4740 train_time:452504ms step_avg:189.73ms
step:2386/4740 train_time:452828ms step_avg:189.79ms
step:2387/4740 train_time:453005ms step_avg:189.78ms
step:2388/4740 train_time:453323ms step_avg:189.83ms
step:2389/4740 train_time:453504ms step_avg:189.83ms
step:2390/4740 train_time:453823ms step_avg:189.88ms
step:2391/4740 train_time:454002ms step_avg:189.88ms
step:2392/4740 train_time:454323ms step_avg:189.93ms
step:2393/4740 train_time:454502ms step_avg:189.93ms
step:2394/4740 train_time:454824ms step_avg:189.99ms
step:2395/4740 train_time:454999ms step_avg:189.98ms
step:2396/4740 train_time:455322ms step_avg:190.03ms
step:2397/4740 train_time:455502ms step_avg:190.03ms
step:2398/4740 train_time:455824ms step_avg:190.09ms
step:2399/4740 train_time:456002ms step_avg:190.08ms
step:2400/4740 train_time:456323ms step_avg:190.13ms
step:2401/4740 train_time:456500ms step_avg:190.13ms
step:2402/4740 train_time:456823ms step_avg:190.18ms
step:2403/4740 train_time:457002ms step_avg:190.18ms
step:2404/4740 train_time:457323ms step_avg:190.23ms
step:2405/4740 train_time:457502ms step_avg:190.23ms
step:2406/4740 train_time:457822ms step_avg:190.28ms
step:2407/4740 train_time:458001ms step_avg:190.28ms
step:2408/4740 train_time:458322ms step_avg:190.33ms
step:2409/4740 train_time:458498ms step_avg:190.33ms
step:2410/4740 train_time:458817ms step_avg:190.38ms
step:2411/4740 train_time:458999ms step_avg:190.38ms
step:2412/4740 train_time:459318ms step_avg:190.43ms
step:2413/4740 train_time:459501ms step_avg:190.43ms
step:2414/4740 train_time:459820ms step_avg:190.48ms
step:2415/4740 train_time:459996ms step_avg:190.47ms
step:2416/4740 train_time:460318ms step_avg:190.53ms
step:2417/4740 train_time:460497ms step_avg:190.52ms
step:2418/4740 train_time:460818ms step_avg:190.58ms
step:2419/4740 train_time:460995ms step_avg:190.57ms
step:2420/4740 train_time:461311ms step_avg:190.62ms
step:2421/4740 train_time:461493ms step_avg:190.62ms
step:2422/4740 train_time:461815ms step_avg:190.68ms
step:2423/4740 train_time:461993ms step_avg:190.67ms
step:2424/4740 train_time:462312ms step_avg:190.72ms
step:2425/4740 train_time:462494ms step_avg:190.72ms
step:2426/4740 train_time:462813ms step_avg:190.77ms
step:2427/4740 train_time:462993ms step_avg:190.77ms
step:2428/4740 train_time:463311ms step_avg:190.82ms
step:2429/4740 train_time:463489ms step_avg:190.81ms
step:2430/4740 train_time:463809ms step_avg:190.87ms
step:2431/4740 train_time:463987ms step_avg:190.86ms
step:2432/4740 train_time:464308ms step_avg:190.92ms
step:2433/4740 train_time:464484ms step_avg:190.91ms
step:2434/4740 train_time:464805ms step_avg:190.96ms
step:2435/4740 train_time:464981ms step_avg:190.96ms
step:2436/4740 train_time:465302ms step_avg:191.01ms
step:2437/4740 train_time:465478ms step_avg:191.00ms
step:2438/4740 train_time:465799ms step_avg:191.06ms
step:2439/4740 train_time:465976ms step_avg:191.05ms
step:2440/4740 train_time:466294ms step_avg:191.10ms
step:2441/4740 train_time:466471ms step_avg:191.10ms
step:2442/4740 train_time:466792ms step_avg:191.15ms
step:2443/4740 train_time:466969ms step_avg:191.15ms
step:2444/4740 train_time:467290ms step_avg:191.20ms
step:2445/4740 train_time:467468ms step_avg:191.19ms
step:2446/4740 train_time:467786ms step_avg:191.25ms
step:2447/4740 train_time:467965ms step_avg:191.24ms
step:2448/4740 train_time:468289ms step_avg:191.29ms
step:2449/4740 train_time:468465ms step_avg:191.29ms
step:2450/4740 train_time:468785ms step_avg:191.34ms
step:2451/4740 train_time:468963ms step_avg:191.34ms
step:2452/4740 train_time:469289ms step_avg:191.39ms
step:2453/4740 train_time:469464ms step_avg:191.38ms
step:2454/4740 train_time:469786ms step_avg:191.44ms
step:2455/4740 train_time:469963ms step_avg:191.43ms
step:2456/4740 train_time:470285ms step_avg:191.48ms
step:2457/4740 train_time:470465ms step_avg:191.48ms
step:2458/4740 train_time:470785ms step_avg:191.53ms
step:2459/4740 train_time:470961ms step_avg:191.53ms
step:2460/4740 train_time:471283ms step_avg:191.58ms
step:2461/4740 train_time:471459ms step_avg:191.57ms
step:2462/4740 train_time:471776ms step_avg:191.62ms
step:2463/4740 train_time:471952ms step_avg:191.62ms
step:2464/4740 train_time:472272ms step_avg:191.67ms
step:2465/4740 train_time:472452ms step_avg:191.66ms
step:2466/4740 train_time:472769ms step_avg:191.71ms
step:2467/4740 train_time:472948ms step_avg:191.71ms
step:2468/4740 train_time:473268ms step_avg:191.76ms
step:2469/4740 train_time:473445ms step_avg:191.76ms
step:2470/4740 train_time:473764ms step_avg:191.81ms
step:2471/4740 train_time:473942ms step_avg:191.80ms
step:2472/4740 train_time:474260ms step_avg:191.85ms
step:2473/4740 train_time:474440ms step_avg:191.85ms
step:2474/4740 train_time:474760ms step_avg:191.90ms
step:2475/4740 train_time:474936ms step_avg:191.89ms
step:2476/4740 train_time:475257ms step_avg:191.95ms
step:2477/4740 train_time:475431ms step_avg:191.94ms
step:2478/4740 train_time:475750ms step_avg:191.99ms
step:2479/4740 train_time:475928ms step_avg:191.98ms
step:2480/4740 train_time:476249ms step_avg:192.04ms
step:2481/4740 train_time:476427ms step_avg:192.03ms
step:2482/4740 train_time:476748ms step_avg:192.08ms
step:2483/4740 train_time:476925ms step_avg:192.08ms
step:2484/4740 train_time:477244ms step_avg:192.13ms
step:2485/4740 train_time:477422ms step_avg:192.12ms
step:2486/4740 train_time:477743ms step_avg:192.17ms
step:2487/4740 train_time:477922ms step_avg:192.17ms
step:2488/4740 train_time:478244ms step_avg:192.22ms
step:2489/4740 train_time:478420ms step_avg:192.21ms
step:2490/4740 train_time:478736ms step_avg:192.26ms
step:2491/4740 train_time:478916ms step_avg:192.26ms
step:2492/4740 train_time:479236ms step_avg:192.31ms
step:2493/4740 train_time:479415ms step_avg:192.30ms
step:2494/4740 train_time:479732ms step_avg:192.35ms
step:2495/4740 train_time:479909ms step_avg:192.35ms
step:2496/4740 train_time:480228ms step_avg:192.40ms
step:2497/4740 train_time:480405ms step_avg:192.39ms
step:2498/4740 train_time:480726ms step_avg:192.44ms
step:2499/4740 train_time:480904ms step_avg:192.44ms
step:2500/4740 train_time:481224ms step_avg:192.49ms
step:2500/4740 val_loss:3.1588 train_time:481348ms step_avg:192.54ms
step:2501/4740 train_time:481406ms step_avg:192.49ms
step:2502/4740 train_time:481738ms step_avg:192.54ms
step:2503/4740 train_time:481915ms step_avg:192.53ms
step:2504/4740 train_time:482230ms step_avg:192.58ms
step:2505/4740 train_time:482404ms step_avg:192.58ms
step:2506/4740 train_time:482731ms step_avg:192.63ms
step:2507/4740 train_time:482911ms step_avg:192.63ms
step:2508/4740 train_time:483229ms step_avg:192.67ms
step:2509/4740 train_time:483405ms step_avg:192.67ms
step:2510/4740 train_time:483728ms step_avg:192.72ms
step:2511/4740 train_time:483905ms step_avg:192.71ms
step:2512/4740 train_time:484223ms step_avg:192.76ms
step:2513/4740 train_time:484399ms step_avg:192.76ms
step:2514/4740 train_time:484719ms step_avg:192.81ms
step:2515/4740 train_time:484897ms step_avg:192.80ms
step:2516/4740 train_time:485217ms step_avg:192.85ms
step:2517/4740 train_time:485393ms step_avg:192.85ms
step:2518/4740 train_time:485711ms step_avg:192.90ms
step:2519/4740 train_time:485890ms step_avg:192.89ms
step:2520/4740 train_time:486210ms step_avg:192.94ms
step:2521/4740 train_time:486388ms step_avg:192.93ms
step:2522/4740 train_time:486708ms step_avg:192.98ms
step:2523/4740 train_time:486882ms step_avg:192.98ms
step:2524/4740 train_time:487206ms step_avg:193.03ms
step:2525/4740 train_time:487382ms step_avg:193.02ms
step:2526/4740 train_time:487702ms step_avg:193.07ms
step:2527/4740 train_time:487878ms step_avg:193.07ms
step:2528/4740 train_time:488199ms step_avg:193.12ms
step:2529/4740 train_time:488375ms step_avg:193.11ms
step:2530/4740 train_time:488696ms step_avg:193.16ms
step:2531/4740 train_time:488874ms step_avg:193.15ms
step:2532/4740 train_time:489197ms step_avg:193.21ms
step:2533/4740 train_time:489376ms step_avg:193.20ms
step:2534/4740 train_time:489699ms step_avg:193.25ms
step:2535/4740 train_time:489878ms step_avg:193.25ms
step:2536/4740 train_time:490198ms step_avg:193.30ms
step:2537/4740 train_time:490377ms step_avg:193.29ms
step:2538/4740 train_time:490699ms step_avg:193.34ms
step:2539/4740 train_time:490874ms step_avg:193.33ms
step:2540/4740 train_time:491196ms step_avg:193.38ms
step:2541/4740 train_time:491375ms step_avg:193.38ms
step:2542/4740 train_time:491696ms step_avg:193.43ms
step:2543/4740 train_time:491872ms step_avg:193.42ms
step:2544/4740 train_time:492191ms step_avg:193.47ms
step:2545/4740 train_time:492369ms step_avg:193.47ms
step:2546/4740 train_time:492687ms step_avg:193.51ms
step:2547/4740 train_time:492866ms step_avg:193.51ms
step:2548/4740 train_time:493185ms step_avg:193.56ms
step:2549/4740 train_time:493359ms step_avg:193.55ms
step:2550/4740 train_time:493680ms step_avg:193.60ms
step:2551/4740 train_time:493858ms step_avg:193.59ms
step:2552/4740 train_time:494177ms step_avg:193.64ms
step:2553/4740 train_time:494354ms step_avg:193.64ms
step:2554/4740 train_time:494676ms step_avg:193.69ms
step:2555/4740 train_time:494852ms step_avg:193.68ms
step:2556/4740 train_time:495172ms step_avg:193.73ms
step:2557/4740 train_time:495353ms step_avg:193.72ms
step:2558/4740 train_time:495677ms step_avg:193.78ms
step:2559/4740 train_time:495854ms step_avg:193.77ms
step:2560/4740 train_time:496172ms step_avg:193.82ms
step:2561/4740 train_time:496351ms step_avg:193.81ms
step:2562/4740 train_time:496671ms step_avg:193.86ms
step:2563/4740 train_time:496849ms step_avg:193.85ms
step:2564/4740 train_time:497167ms step_avg:193.90ms
step:2565/4740 train_time:497346ms step_avg:193.90ms
step:2566/4740 train_time:497666ms step_avg:193.95ms
step:2567/4740 train_time:497846ms step_avg:193.94ms
step:2568/4740 train_time:498165ms step_avg:193.99ms
step:2569/4740 train_time:498341ms step_avg:193.98ms
step:2570/4740 train_time:498660ms step_avg:194.03ms
step:2571/4740 train_time:498841ms step_avg:194.03ms
step:2572/4740 train_time:499161ms step_avg:194.08ms
step:2573/4740 train_time:499338ms step_avg:194.07ms
step:2574/4740 train_time:499660ms step_avg:194.12ms
step:2575/4740 train_time:499838ms step_avg:194.11ms
step:2576/4740 train_time:500157ms step_avg:194.16ms
step:2577/4740 train_time:500336ms step_avg:194.15ms
step:2578/4740 train_time:500656ms step_avg:194.20ms
step:2579/4740 train_time:500834ms step_avg:194.20ms
step:2580/4740 train_time:501156ms step_avg:194.25ms
step:2581/4740 train_time:501331ms step_avg:194.24ms
step:2582/4740 train_time:501648ms step_avg:194.29ms
step:2583/4740 train_time:501825ms step_avg:194.28ms
step:2584/4740 train_time:502144ms step_avg:194.33ms
step:2585/4740 train_time:502322ms step_avg:194.32ms
step:2586/4740 train_time:502641ms step_avg:194.37ms
step:2587/4740 train_time:502818ms step_avg:194.36ms
step:2588/4740 train_time:503141ms step_avg:194.41ms
step:2589/4740 train_time:503318ms step_avg:194.41ms
step:2590/4740 train_time:503641ms step_avg:194.46ms
step:2591/4740 train_time:503816ms step_avg:194.45ms
step:2592/4740 train_time:504136ms step_avg:194.50ms
step:2593/4740 train_time:504314ms step_avg:194.49ms
step:2594/4740 train_time:504633ms step_avg:194.54ms
step:2595/4740 train_time:504815ms step_avg:194.53ms
step:2596/4740 train_time:505136ms step_avg:194.58ms
step:2597/4740 train_time:505313ms step_avg:194.58ms
step:2598/4740 train_time:505635ms step_avg:194.62ms
step:2599/4740 train_time:505815ms step_avg:194.62ms
step:2600/4740 train_time:506135ms step_avg:194.67ms
step:2601/4740 train_time:506312ms step_avg:194.66ms
step:2602/4740 train_time:506630ms step_avg:194.71ms
step:2603/4740 train_time:506809ms step_avg:194.70ms
step:2604/4740 train_time:507130ms step_avg:194.75ms
step:2605/4740 train_time:507307ms step_avg:194.74ms
step:2606/4740 train_time:507625ms step_avg:194.79ms
step:2607/4740 train_time:507805ms step_avg:194.79ms
step:2608/4740 train_time:508124ms step_avg:194.83ms
step:2609/4740 train_time:508302ms step_avg:194.83ms
step:2610/4740 train_time:508623ms step_avg:194.87ms
step:2611/4740 train_time:508800ms step_avg:194.87ms
step:2612/4740 train_time:509120ms step_avg:194.92ms
step:2613/4740 train_time:509301ms step_avg:194.91ms
step:2614/4740 train_time:509620ms step_avg:194.96ms
step:2615/4740 train_time:509799ms step_avg:194.95ms
step:2616/4740 train_time:510122ms step_avg:195.00ms
step:2617/4740 train_time:510300ms step_avg:194.99ms
step:2618/4740 train_time:510622ms step_avg:195.04ms
step:2619/4740 train_time:510800ms step_avg:195.04ms
step:2620/4740 train_time:511121ms step_avg:195.08ms
step:2621/4740 train_time:511297ms step_avg:195.08ms
step:2622/4740 train_time:511619ms step_avg:195.13ms
step:2623/4740 train_time:511795ms step_avg:195.12ms
step:2624/4740 train_time:512116ms step_avg:195.17ms
step:2625/4740 train_time:512292ms step_avg:195.16ms
step:2626/4740 train_time:512610ms step_avg:195.21ms
step:2627/4740 train_time:512787ms step_avg:195.20ms
step:2628/4740 train_time:513108ms step_avg:195.25ms
step:2629/4740 train_time:513285ms step_avg:195.24ms
step:2630/4740 train_time:513604ms step_avg:195.29ms
step:2631/4740 train_time:513784ms step_avg:195.28ms
step:2632/4740 train_time:514102ms step_avg:195.33ms
step:2633/4740 train_time:514279ms step_avg:195.32ms
step:2634/4740 train_time:514599ms step_avg:195.37ms
step:2635/4740 train_time:514778ms step_avg:195.36ms
step:2636/4740 train_time:515100ms step_avg:195.41ms
step:2637/4740 train_time:515276ms step_avg:195.40ms
step:2638/4740 train_time:515597ms step_avg:195.45ms
step:2639/4740 train_time:515772ms step_avg:195.44ms
step:2640/4740 train_time:516091ms step_avg:195.49ms
step:2641/4740 train_time:516271ms step_avg:195.48ms
step:2642/4740 train_time:516592ms step_avg:195.53ms
step:2643/4740 train_time:516768ms step_avg:195.52ms
step:2644/4740 train_time:517086ms step_avg:195.57ms
step:2645/4740 train_time:517263ms step_avg:195.56ms
step:2646/4740 train_time:517583ms step_avg:195.61ms
step:2647/4740 train_time:517762ms step_avg:195.60ms
step:2648/4740 train_time:518081ms step_avg:195.65ms
step:2649/4740 train_time:518258ms step_avg:195.64ms
step:2650/4740 train_time:518581ms step_avg:195.69ms
step:2651/4740 train_time:518757ms step_avg:195.68ms
step:2652/4740 train_time:519078ms step_avg:195.73ms
step:2653/4740 train_time:519253ms step_avg:195.72ms
step:2654/4740 train_time:519570ms step_avg:195.77ms
step:2655/4740 train_time:519750ms step_avg:195.76ms
step:2656/4740 train_time:520069ms step_avg:195.81ms
step:2657/4740 train_time:520249ms step_avg:195.80ms
step:2658/4740 train_time:520566ms step_avg:195.85ms
step:2659/4740 train_time:520750ms step_avg:195.84ms
step:2660/4740 train_time:521069ms step_avg:195.89ms
step:2661/4740 train_time:521245ms step_avg:195.88ms
step:2662/4740 train_time:521566ms step_avg:195.93ms
step:2663/4740 train_time:521742ms step_avg:195.92ms
step:2664/4740 train_time:522060ms step_avg:195.97ms
step:2665/4740 train_time:522239ms step_avg:195.96ms
step:2666/4740 train_time:522563ms step_avg:196.01ms
step:2667/4740 train_time:522740ms step_avg:196.00ms
step:2668/4740 train_time:523058ms step_avg:196.05ms
step:2669/4740 train_time:523235ms step_avg:196.04ms
step:2670/4740 train_time:523555ms step_avg:196.09ms
step:2671/4740 train_time:523732ms step_avg:196.08ms
step:2672/4740 train_time:524050ms step_avg:196.13ms
step:2673/4740 train_time:524229ms step_avg:196.12ms
step:2674/4740 train_time:524552ms step_avg:196.17ms
step:2675/4740 train_time:524728ms step_avg:196.16ms
step:2676/4740 train_time:525048ms step_avg:196.21ms
step:2677/4740 train_time:525229ms step_avg:196.20ms
step:2678/4740 train_time:525549ms step_avg:196.25ms
step:2679/4740 train_time:525728ms step_avg:196.24ms
step:2680/4740 train_time:526048ms step_avg:196.29ms
step:2681/4740 train_time:526225ms step_avg:196.28ms
step:2682/4740 train_time:526547ms step_avg:196.33ms
step:2683/4740 train_time:526725ms step_avg:196.32ms
step:2684/4740 train_time:527046ms step_avg:196.37ms
step:2685/4740 train_time:527221ms step_avg:196.36ms
step:2686/4740 train_time:527541ms step_avg:196.40ms
step:2687/4740 train_time:527719ms step_avg:196.40ms
step:2688/4740 train_time:528040ms step_avg:196.44ms
step:2689/4740 train_time:528218ms step_avg:196.44ms
step:2690/4740 train_time:528540ms step_avg:196.48ms
step:2691/4740 train_time:528714ms step_avg:196.47ms
step:2692/4740 train_time:529034ms step_avg:196.52ms
step:2693/4740 train_time:529211ms step_avg:196.51ms
step:2694/4740 train_time:529532ms step_avg:196.56ms
step:2695/4740 train_time:529711ms step_avg:196.55ms
step:2696/4740 train_time:530029ms step_avg:196.60ms
step:2697/4740 train_time:530208ms step_avg:196.59ms
step:2698/4740 train_time:530526ms step_avg:196.64ms
step:2699/4740 train_time:530703ms step_avg:196.63ms
step:2700/4740 train_time:531019ms step_avg:196.67ms
step:2701/4740 train_time:531200ms step_avg:196.67ms
step:2702/4740 train_time:531521ms step_avg:196.71ms
step:2703/4740 train_time:531699ms step_avg:196.71ms
step:2704/4740 train_time:532017ms step_avg:196.75ms
step:2705/4740 train_time:532197ms step_avg:196.75ms
step:2706/4740 train_time:532518ms step_avg:196.79ms
step:2707/4740 train_time:532697ms step_avg:196.78ms
step:2708/4740 train_time:533017ms step_avg:196.83ms
step:2709/4740 train_time:533194ms step_avg:196.82ms
step:2710/4740 train_time:533517ms step_avg:196.87ms
step:2711/4740 train_time:533694ms step_avg:196.86ms
step:2712/4740 train_time:534015ms step_avg:196.91ms
step:2713/4740 train_time:534192ms step_avg:196.90ms
step:2714/4740 train_time:534507ms step_avg:196.94ms
step:2715/4740 train_time:534690ms step_avg:196.94ms
step:2716/4740 train_time:535007ms step_avg:196.98ms
step:2717/4740 train_time:535187ms step_avg:196.98ms
step:2718/4740 train_time:535507ms step_avg:197.02ms
step:2719/4740 train_time:535684ms step_avg:197.02ms
step:2720/4740 train_time:536002ms step_avg:197.06ms
step:2721/4740 train_time:536179ms step_avg:197.05ms
step:2722/4740 train_time:536500ms step_avg:197.10ms
step:2723/4740 train_time:536678ms step_avg:197.09ms
step:2724/4740 train_time:537000ms step_avg:197.14ms
step:2725/4740 train_time:537175ms step_avg:197.13ms
step:2726/4740 train_time:537493ms step_avg:197.17ms
step:2727/4740 train_time:537673ms step_avg:197.17ms
step:2728/4740 train_time:537992ms step_avg:197.21ms
step:2729/4740 train_time:538172ms step_avg:197.20ms
step:2730/4740 train_time:538492ms step_avg:197.25ms
step:2731/4740 train_time:538672ms step_avg:197.24ms
step:2732/4740 train_time:538992ms step_avg:197.29ms
step:2733/4740 train_time:539169ms step_avg:197.28ms
step:2734/4740 train_time:539489ms step_avg:197.33ms
step:2735/4740 train_time:539664ms step_avg:197.32ms
step:2736/4740 train_time:539983ms step_avg:197.36ms
step:2737/4740 train_time:540161ms step_avg:197.36ms
step:2738/4740 train_time:540482ms step_avg:197.40ms
step:2739/4740 train_time:540660ms step_avg:197.39ms
step:2740/4740 train_time:540980ms step_avg:197.44ms
step:2741/4740 train_time:541157ms step_avg:197.43ms
step:2742/4740 train_time:541477ms step_avg:197.48ms
step:2743/4740 train_time:541655ms step_avg:197.47ms
step:2744/4740 train_time:541979ms step_avg:197.51ms
step:2745/4740 train_time:542154ms step_avg:197.51ms
step:2746/4740 train_time:542475ms step_avg:197.55ms
step:2747/4740 train_time:542651ms step_avg:197.54ms
step:2748/4740 train_time:542969ms step_avg:197.59ms
step:2749/4740 train_time:543149ms step_avg:197.58ms
step:2750/4740 train_time:543467ms step_avg:197.62ms
step:2750/4740 val_loss:3.1271 train_time:543592ms step_avg:197.67ms
step:2751/4740 train_time:543650ms step_avg:197.62ms
step:2752/4740 train_time:543985ms step_avg:197.67ms
step:2753/4740 train_time:544159ms step_avg:197.66ms
step:2754/4740 train_time:544472ms step_avg:197.70ms
step:2755/4740 train_time:544648ms step_avg:197.69ms
step:2756/4740 train_time:544976ms step_avg:197.74ms
step:2757/4740 train_time:545156ms step_avg:197.74ms
step:2758/4740 train_time:545474ms step_avg:197.78ms
step:2759/4740 train_time:545650ms step_avg:197.77ms
step:2760/4740 train_time:545974ms step_avg:197.82ms
step:2761/4740 train_time:546152ms step_avg:197.81ms
step:2762/4740 train_time:546476ms step_avg:197.86ms
step:2763/4740 train_time:546655ms step_avg:197.85ms
step:2764/4740 train_time:546977ms step_avg:197.89ms
step:2765/4740 train_time:547154ms step_avg:197.89ms
step:2766/4740 train_time:547476ms step_avg:197.93ms
step:2767/4740 train_time:547652ms step_avg:197.92ms
step:2768/4740 train_time:547974ms step_avg:197.97ms
step:2769/4740 train_time:548152ms step_avg:197.96ms
step:2770/4740 train_time:548474ms step_avg:198.01ms
step:2771/4740 train_time:548652ms step_avg:198.00ms
step:2772/4740 train_time:548972ms step_avg:198.04ms
step:2773/4740 train_time:549149ms step_avg:198.03ms
step:2774/4740 train_time:549474ms step_avg:198.08ms
step:2775/4740 train_time:549650ms step_avg:198.07ms
step:2776/4740 train_time:549971ms step_avg:198.12ms
step:2777/4740 train_time:550148ms step_avg:198.11ms
step:2778/4740 train_time:550471ms step_avg:198.15ms
step:2779/4740 train_time:550646ms step_avg:198.15ms
step:2780/4740 train_time:550969ms step_avg:198.19ms
step:2781/4740 train_time:551146ms step_avg:198.18ms
step:2782/4740 train_time:551468ms step_avg:198.23ms
step:2783/4740 train_time:551648ms step_avg:198.22ms
step:2784/4740 train_time:551969ms step_avg:198.26ms
step:2785/4740 train_time:552145ms step_avg:198.26ms
step:2786/4740 train_time:552467ms step_avg:198.30ms
step:2787/4740 train_time:552643ms step_avg:198.29ms
step:2788/4740 train_time:552962ms step_avg:198.34ms
step:2789/4740 train_time:553141ms step_avg:198.33ms
step:2790/4740 train_time:553462ms step_avg:198.37ms
step:2791/4740 train_time:553638ms step_avg:198.37ms
step:2792/4740 train_time:553959ms step_avg:198.41ms
step:2793/4740 train_time:554140ms step_avg:198.40ms
step:2794/4740 train_time:554462ms step_avg:198.45ms
step:2795/4740 train_time:554639ms step_avg:198.44ms
step:2796/4740 train_time:554960ms step_avg:198.48ms
step:2797/4740 train_time:555140ms step_avg:198.48ms
step:2798/4740 train_time:555457ms step_avg:198.52ms
step:2799/4740 train_time:555636ms step_avg:198.51ms
step:2800/4740 train_time:555958ms step_avg:198.56ms
step:2801/4740 train_time:556136ms step_avg:198.55ms
step:2802/4740 train_time:556454ms step_avg:198.59ms
step:2803/4740 train_time:556634ms step_avg:198.58ms
step:2804/4740 train_time:556956ms step_avg:198.63ms
step:2805/4740 train_time:557135ms step_avg:198.62ms
step:2806/4740 train_time:557453ms step_avg:198.66ms
step:2807/4740 train_time:557636ms step_avg:198.66ms
step:2808/4740 train_time:557959ms step_avg:198.70ms
step:2809/4740 train_time:558139ms step_avg:198.70ms
step:2810/4740 train_time:558464ms step_avg:198.74ms
step:2811/4740 train_time:558642ms step_avg:198.73ms
step:2812/4740 train_time:558959ms step_avg:198.78ms
step:2813/4740 train_time:559140ms step_avg:198.77ms
step:2814/4740 train_time:559465ms step_avg:198.81ms
step:2815/4740 train_time:559639ms step_avg:198.81ms
step:2816/4740 train_time:559957ms step_avg:198.85ms
step:2817/4740 train_time:560134ms step_avg:198.84ms
step:2818/4740 train_time:560453ms step_avg:198.88ms
step:2819/4740 train_time:560633ms step_avg:198.88ms
step:2820/4740 train_time:560952ms step_avg:198.92ms
step:2821/4740 train_time:561131ms step_avg:198.91ms
step:2822/4740 train_time:561455ms step_avg:198.96ms
step:2823/4740 train_time:561633ms step_avg:198.95ms
step:2824/4740 train_time:561953ms step_avg:198.99ms
step:2825/4740 train_time:562131ms step_avg:198.98ms
step:2826/4740 train_time:562449ms step_avg:199.03ms
step:2827/4740 train_time:562629ms step_avg:199.02ms
step:2828/4740 train_time:562948ms step_avg:199.06ms
step:2829/4740 train_time:563129ms step_avg:199.06ms
step:2830/4740 train_time:563450ms step_avg:199.10ms
step:2831/4740 train_time:563627ms step_avg:199.09ms
step:2832/4740 train_time:563946ms step_avg:199.13ms
step:2833/4740 train_time:564123ms step_avg:199.13ms
step:2834/4740 train_time:564443ms step_avg:199.17ms
step:2835/4740 train_time:564622ms step_avg:199.16ms
step:2836/4740 train_time:564943ms step_avg:199.20ms
step:2837/4740 train_time:565122ms step_avg:199.20ms
step:2838/4740 train_time:565441ms step_avg:199.24ms
step:2839/4740 train_time:565619ms step_avg:199.23ms
step:2840/4740 train_time:565938ms step_avg:199.27ms
step:2841/4740 train_time:566117ms step_avg:199.27ms
step:2842/4740 train_time:566437ms step_avg:199.31ms
step:2843/4740 train_time:566618ms step_avg:199.30ms
step:2844/4740 train_time:566935ms step_avg:199.34ms
step:2845/4740 train_time:567116ms step_avg:199.34ms
step:2846/4740 train_time:567435ms step_avg:199.38ms
step:2847/4740 train_time:567614ms step_avg:199.37ms
step:2848/4740 train_time:567936ms step_avg:199.42ms
step:2849/4740 train_time:568113ms step_avg:199.41ms
step:2850/4740 train_time:568435ms step_avg:199.45ms
step:2851/4740 train_time:568616ms step_avg:199.44ms
step:2852/4740 train_time:568934ms step_avg:199.49ms
step:2853/4740 train_time:569112ms step_avg:199.48ms
step:2854/4740 train_time:569436ms step_avg:199.52ms
step:2855/4740 train_time:569612ms step_avg:199.51ms
step:2856/4740 train_time:569928ms step_avg:199.55ms
step:2857/4740 train_time:570108ms step_avg:199.55ms
step:2858/4740 train_time:570431ms step_avg:199.59ms
step:2859/4740 train_time:570610ms step_avg:199.58ms
step:2860/4740 train_time:570930ms step_avg:199.63ms
step:2861/4740 train_time:571108ms step_avg:199.62ms
step:2862/4740 train_time:571428ms step_avg:199.66ms
step:2863/4740 train_time:571607ms step_avg:199.65ms
step:2864/4740 train_time:571926ms step_avg:199.69ms
step:2865/4740 train_time:572107ms step_avg:199.69ms
step:2866/4740 train_time:572428ms step_avg:199.73ms
step:2867/4740 train_time:572605ms step_avg:199.72ms
step:2868/4740 train_time:572924ms step_avg:199.76ms
step:2869/4740 train_time:573101ms step_avg:199.76ms
step:2870/4740 train_time:573426ms step_avg:199.80ms
step:2871/4740 train_time:573603ms step_avg:199.79ms
step:2872/4740 train_time:573925ms step_avg:199.83ms
step:2873/4740 train_time:574105ms step_avg:199.83ms
step:2874/4740 train_time:574426ms step_avg:199.87ms
step:2875/4740 train_time:574603ms step_avg:199.86ms
step:2876/4740 train_time:574924ms step_avg:199.90ms
step:2877/4740 train_time:575101ms step_avg:199.90ms
step:2878/4740 train_time:575419ms step_avg:199.94ms
step:2879/4740 train_time:575601ms step_avg:199.93ms
step:2880/4740 train_time:575921ms step_avg:199.97ms
step:2881/4740 train_time:576098ms step_avg:199.96ms
step:2882/4740 train_time:576417ms step_avg:200.01ms
step:2883/4740 train_time:576594ms step_avg:200.00ms
step:2884/4740 train_time:576913ms step_avg:200.04ms
step:2885/4740 train_time:577094ms step_avg:200.03ms
step:2886/4740 train_time:577415ms step_avg:200.07ms
step:2887/4740 train_time:577592ms step_avg:200.07ms
step:2888/4740 train_time:577912ms step_avg:200.11ms
step:2889/4740 train_time:578091ms step_avg:200.10ms
step:2890/4740 train_time:578411ms step_avg:200.14ms
step:2891/4740 train_time:578591ms step_avg:200.14ms
step:2892/4740 train_time:578912ms step_avg:200.18ms
step:2893/4740 train_time:579091ms step_avg:200.17ms
step:2894/4740 train_time:579414ms step_avg:200.21ms
step:2895/4740 train_time:579595ms step_avg:200.21ms
step:2896/4740 train_time:579914ms step_avg:200.25ms
step:2897/4740 train_time:580090ms step_avg:200.24ms
step:2898/4740 train_time:580411ms step_avg:200.28ms
step:2899/4740 train_time:580592ms step_avg:200.27ms
step:2900/4740 train_time:580913ms step_avg:200.31ms
step:2901/4740 train_time:581091ms step_avg:200.31ms
step:2902/4740 train_time:581414ms step_avg:200.35ms
step:2903/4740 train_time:581594ms step_avg:200.34ms
step:2904/4740 train_time:581914ms step_avg:200.38ms
step:2905/4740 train_time:582091ms step_avg:200.38ms
step:2906/4740 train_time:582411ms step_avg:200.42ms
step:2907/4740 train_time:582588ms step_avg:200.41ms
step:2908/4740 train_time:582912ms step_avg:200.45ms
step:2909/4740 train_time:583088ms step_avg:200.44ms
step:2910/4740 train_time:583409ms step_avg:200.48ms
step:2911/4740 train_time:583590ms step_avg:200.48ms
step:2912/4740 train_time:583908ms step_avg:200.52ms
step:2913/4740 train_time:584087ms step_avg:200.51ms
step:2914/4740 train_time:584407ms step_avg:200.55ms
step:2915/4740 train_time:584589ms step_avg:200.55ms
step:2916/4740 train_time:584909ms step_avg:200.59ms
step:2917/4740 train_time:585088ms step_avg:200.58ms
step:2918/4740 train_time:585407ms step_avg:200.62ms
step:2919/4740 train_time:585589ms step_avg:200.61ms
step:2920/4740 train_time:585908ms step_avg:200.65ms
step:2921/4740 train_time:586086ms step_avg:200.65ms
step:2922/4740 train_time:586407ms step_avg:200.69ms
step:2923/4740 train_time:586585ms step_avg:200.68ms
step:2924/4740 train_time:586905ms step_avg:200.72ms
step:2925/4740 train_time:587083ms step_avg:200.71ms
step:2926/4740 train_time:587406ms step_avg:200.75ms
step:2927/4740 train_time:587583ms step_avg:200.75ms
step:2928/4740 train_time:587905ms step_avg:200.79ms
step:2929/4740 train_time:588083ms step_avg:200.78ms
step:2930/4740 train_time:588403ms step_avg:200.82ms
step:2931/4740 train_time:588582ms step_avg:200.81ms
step:2932/4740 train_time:588904ms step_avg:200.85ms
step:2933/4740 train_time:589085ms step_avg:200.85ms
step:2934/4740 train_time:589404ms step_avg:200.89ms
step:2935/4740 train_time:589585ms step_avg:200.88ms
step:2936/4740 train_time:589905ms step_avg:200.92ms
step:2937/4740 train_time:590086ms step_avg:200.91ms
step:2938/4740 train_time:590404ms step_avg:200.95ms
step:2939/4740 train_time:590584ms step_avg:200.95ms
step:2940/4740 train_time:590904ms step_avg:200.99ms
step:2941/4740 train_time:591081ms step_avg:200.98ms
step:2942/4740 train_time:591402ms step_avg:201.02ms
step:2943/4740 train_time:591584ms step_avg:201.01ms
step:2944/4740 train_time:591905ms step_avg:201.05ms
step:2945/4740 train_time:592081ms step_avg:201.05ms
step:2946/4740 train_time:592400ms step_avg:201.09ms
step:2947/4740 train_time:592578ms step_avg:201.08ms
step:2948/4740 train_time:592900ms step_avg:201.12ms
step:2949/4740 train_time:593081ms step_avg:201.11ms
step:2950/4740 train_time:593400ms step_avg:201.15ms
step:2951/4740 train_time:593578ms step_avg:201.14ms
step:2952/4740 train_time:593896ms step_avg:201.18ms
step:2953/4740 train_time:594073ms step_avg:201.18ms
step:2954/4740 train_time:594391ms step_avg:201.22ms
step:2955/4740 train_time:594571ms step_avg:201.21ms
step:2956/4740 train_time:594892ms step_avg:201.25ms
step:2957/4740 train_time:595068ms step_avg:201.24ms
step:2958/4740 train_time:595391ms step_avg:201.28ms
step:2959/4740 train_time:595569ms step_avg:201.27ms
step:2960/4740 train_time:595890ms step_avg:201.31ms
step:2961/4740 train_time:596066ms step_avg:201.31ms
step:2962/4740 train_time:596388ms step_avg:201.35ms
step:2963/4740 train_time:596564ms step_avg:201.34ms
step:2964/4740 train_time:596884ms step_avg:201.38ms
step:2965/4740 train_time:597062ms step_avg:201.37ms
step:2966/4740 train_time:597380ms step_avg:201.41ms
step:2967/4740 train_time:597558ms step_avg:201.40ms
step:2968/4740 train_time:597878ms step_avg:201.44ms
step:2969/4740 train_time:598061ms step_avg:201.44ms
step:2970/4740 train_time:598382ms step_avg:201.48ms
step:2971/4740 train_time:598562ms step_avg:201.47ms
step:2972/4740 train_time:598879ms step_avg:201.51ms
step:2973/4740 train_time:599057ms step_avg:201.50ms
step:2974/4740 train_time:599377ms step_avg:201.54ms
step:2975/4740 train_time:599553ms step_avg:201.53ms
step:2976/4740 train_time:599876ms step_avg:201.57ms
step:2977/4740 train_time:600054ms step_avg:201.56ms
step:2978/4740 train_time:600373ms step_avg:201.60ms
step:2979/4740 train_time:600552ms step_avg:201.60ms
step:2980/4740 train_time:600875ms step_avg:201.64ms
step:2981/4740 train_time:601052ms step_avg:201.63ms
step:2982/4740 train_time:601375ms step_avg:201.67ms
step:2983/4740 train_time:601552ms step_avg:201.66ms
step:2984/4740 train_time:601871ms step_avg:201.70ms
step:2985/4740 train_time:602049ms step_avg:201.69ms
step:2986/4740 train_time:602373ms step_avg:201.73ms
step:2987/4740 train_time:602549ms step_avg:201.72ms
step:2988/4740 train_time:602871ms step_avg:201.76ms
step:2989/4740 train_time:603049ms step_avg:201.76ms
step:2990/4740 train_time:603371ms step_avg:201.80ms
step:2991/4740 train_time:603546ms step_avg:201.79ms
step:2992/4740 train_time:603867ms step_avg:201.83ms
step:2993/4740 train_time:604043ms step_avg:201.82ms
step:2994/4740 train_time:604366ms step_avg:201.86ms
step:2995/4740 train_time:604543ms step_avg:201.85ms
step:2996/4740 train_time:604865ms step_avg:201.89ms
step:2997/4740 train_time:605041ms step_avg:201.88ms
step:2998/4740 train_time:605364ms step_avg:201.92ms
step:2999/4740 train_time:605540ms step_avg:201.91ms
step:3000/4740 train_time:605862ms step_avg:201.95ms
step:3000/4740 val_loss:3.0970 train_time:605983ms step_avg:201.99ms
step:3001/4740 train_time:606041ms step_avg:201.95ms
step:3002/4740 train_time:606376ms step_avg:201.99ms
step:3003/4740 train_time:606552ms step_avg:201.98ms
step:3004/4740 train_time:606865ms step_avg:202.02ms
step:3005/4740 train_time:607045ms step_avg:202.01ms
step:3006/4740 train_time:607370ms step_avg:202.05ms
step:3007/4740 train_time:607550ms step_avg:202.05ms
step:3008/4740 train_time:607870ms step_avg:202.08ms
step:3009/4740 train_time:608046ms step_avg:202.08ms
step:3010/4740 train_time:608367ms step_avg:202.12ms
step:3011/4740 train_time:608549ms step_avg:202.11ms
step:3012/4740 train_time:608867ms step_avg:202.15ms
step:3013/4740 train_time:609043ms step_avg:202.14ms
step:3014/4740 train_time:609364ms step_avg:202.18ms
step:3015/4740 train_time:609544ms step_avg:202.17ms
step:3016/4740 train_time:609863ms step_avg:202.21ms
step:3017/4740 train_time:610041ms step_avg:202.20ms
step:3018/4740 train_time:610361ms step_avg:202.24ms
step:3019/4740 train_time:610539ms step_avg:202.23ms
step:3020/4740 train_time:610862ms step_avg:202.27ms
step:3021/4740 train_time:611038ms step_avg:202.26ms
step:3022/4740 train_time:611359ms step_avg:202.30ms
step:3023/4740 train_time:611536ms step_avg:202.29ms
step:3024/4740 train_time:611857ms step_avg:202.33ms
step:3025/4740 train_time:612036ms step_avg:202.33ms
step:3026/4740 train_time:612355ms step_avg:202.36ms
step:3027/4740 train_time:612540ms step_avg:202.36ms
step:3028/4740 train_time:612857ms step_avg:202.40ms
step:3029/4740 train_time:613036ms step_avg:202.39ms
step:3030/4740 train_time:613357ms step_avg:202.43ms
step:3031/4740 train_time:613537ms step_avg:202.42ms
step:3032/4740 train_time:613859ms step_avg:202.46ms
step:3033/4740 train_time:614034ms step_avg:202.45ms
step:3034/4740 train_time:614356ms step_avg:202.49ms
step:3035/4740 train_time:614533ms step_avg:202.48ms
step:3036/4740 train_time:614856ms step_avg:202.52ms
step:3037/4740 train_time:615036ms step_avg:202.51ms
step:3038/4740 train_time:615357ms step_avg:202.55ms
step:3039/4740 train_time:615535ms step_avg:202.55ms
step:3040/4740 train_time:615860ms step_avg:202.59ms
step:3041/4740 train_time:616037ms step_avg:202.58ms
step:3042/4740 train_time:616358ms step_avg:202.62ms
step:3043/4740 train_time:616534ms step_avg:202.61ms
step:3044/4740 train_time:616855ms step_avg:202.65ms
step:3045/4740 train_time:617034ms step_avg:202.64ms
step:3046/4740 train_time:617356ms step_avg:202.68ms
step:3047/4740 train_time:617530ms step_avg:202.67ms
step:3048/4740 train_time:617854ms step_avg:202.71ms
step:3049/4740 train_time:618030ms step_avg:202.70ms
step:3050/4740 train_time:618352ms step_avg:202.74ms
step:3051/4740 train_time:618530ms step_avg:202.73ms
step:3052/4740 train_time:618848ms step_avg:202.77ms
step:3053/4740 train_time:619028ms step_avg:202.76ms
step:3054/4740 train_time:619350ms step_avg:202.80ms
step:3055/4740 train_time:619528ms step_avg:202.79ms
step:3056/4740 train_time:619851ms step_avg:202.83ms
step:3057/4740 train_time:620029ms step_avg:202.82ms
step:3058/4740 train_time:620348ms step_avg:202.86ms
step:3059/4740 train_time:620525ms step_avg:202.85ms
step:3060/4740 train_time:620844ms step_avg:202.89ms
step:3061/4740 train_time:621020ms step_avg:202.88ms
step:3062/4740 train_time:621339ms step_avg:202.92ms
step:3063/4740 train_time:621516ms step_avg:202.91ms
step:3064/4740 train_time:621836ms step_avg:202.95ms
step:3065/4740 train_time:622012ms step_avg:202.94ms
step:3066/4740 train_time:622332ms step_avg:202.98ms
step:3067/4740 train_time:622510ms step_avg:202.97ms
step:3068/4740 train_time:622827ms step_avg:203.01ms
step:3069/4740 train_time:623008ms step_avg:203.00ms
step:3070/4740 train_time:623327ms step_avg:203.04ms
step:3071/4740 train_time:623506ms step_avg:203.03ms
step:3072/4740 train_time:623825ms step_avg:203.07ms
step:3073/4740 train_time:624002ms step_avg:203.06ms
step:3074/4740 train_time:624320ms step_avg:203.10ms
step:3075/4740 train_time:624499ms step_avg:203.09ms
step:3076/4740 train_time:624816ms step_avg:203.13ms
step:3077/4740 train_time:624999ms step_avg:203.12ms
step:3078/4740 train_time:625318ms step_avg:203.16ms
step:3079/4740 train_time:625498ms step_avg:203.15ms
step:3080/4740 train_time:625816ms step_avg:203.19ms
step:3081/4740 train_time:625999ms step_avg:203.18ms
step:3082/4740 train_time:626322ms step_avg:203.22ms
step:3083/4740 train_time:626497ms step_avg:203.21ms
step:3084/4740 train_time:626820ms step_avg:203.25ms
step:3085/4740 train_time:626997ms step_avg:203.24ms
step:3086/4740 train_time:627317ms step_avg:203.28ms
step:3087/4740 train_time:627496ms step_avg:203.27ms
step:3088/4740 train_time:627818ms step_avg:203.31ms
step:3089/4740 train_time:627997ms step_avg:203.30ms
step:3090/4740 train_time:628317ms step_avg:203.34ms
step:3091/4740 train_time:628498ms step_avg:203.33ms
step:3092/4740 train_time:628821ms step_avg:203.37ms
step:3093/4740 train_time:628996ms step_avg:203.36ms
step:3094/4740 train_time:629316ms step_avg:203.40ms
step:3095/4740 train_time:629494ms step_avg:203.39ms
step:3096/4740 train_time:629814ms step_avg:203.43ms
step:3097/4740 train_time:629991ms step_avg:203.42ms
step:3098/4740 train_time:630313ms step_avg:203.46ms
step:3099/4740 train_time:630493ms step_avg:203.45ms
step:3100/4740 train_time:630815ms step_avg:203.49ms
step:3101/4740 train_time:630993ms step_avg:203.48ms
step:3102/4740 train_time:631315ms step_avg:203.52ms
step:3103/4740 train_time:631494ms step_avg:203.51ms
step:3104/4740 train_time:631817ms step_avg:203.55ms
step:3105/4740 train_time:631996ms step_avg:203.54ms
step:3106/4740 train_time:632315ms step_avg:203.58ms
step:3107/4740 train_time:632494ms step_avg:203.57ms
step:3108/4740 train_time:632817ms step_avg:203.61ms
step:3109/4740 train_time:632993ms step_avg:203.60ms
step:3110/4740 train_time:633314ms step_avg:203.64ms
step:3111/4740 train_time:633493ms step_avg:203.63ms
step:3112/4740 train_time:633814ms step_avg:203.67ms
step:3113/4740 train_time:633990ms step_avg:203.66ms
step:3114/4740 train_time:634306ms step_avg:203.69ms
step:3115/4740 train_time:634485ms step_avg:203.69ms
step:3116/4740 train_time:634806ms step_avg:203.72ms
step:3117/4740 train_time:634984ms step_avg:203.72ms
step:3118/4740 train_time:635303ms step_avg:203.75ms
step:3119/4740 train_time:635480ms step_avg:203.74ms
step:3120/4740 train_time:635799ms step_avg:203.78ms
step:3121/4740 train_time:635978ms step_avg:203.77ms
step:3122/4740 train_time:636301ms step_avg:203.81ms
step:3123/4740 train_time:636479ms step_avg:203.80ms
step:3124/4740 train_time:636799ms step_avg:203.84ms
step:3125/4740 train_time:636977ms step_avg:203.83ms
step:3126/4740 train_time:637299ms step_avg:203.87ms
step:3127/4740 train_time:637477ms step_avg:203.86ms
step:3128/4740 train_time:637800ms step_avg:203.90ms
step:3129/4740 train_time:637977ms step_avg:203.89ms
step:3130/4740 train_time:638299ms step_avg:203.93ms
step:3131/4740 train_time:638477ms step_avg:203.92ms
step:3132/4740 train_time:638800ms step_avg:203.96ms
step:3133/4740 train_time:638976ms step_avg:203.95ms
step:3134/4740 train_time:639301ms step_avg:203.99ms
step:3135/4740 train_time:639477ms step_avg:203.98ms
step:3136/4740 train_time:639800ms step_avg:204.02ms
step:3137/4740 train_time:639977ms step_avg:204.01ms
step:3138/4740 train_time:640301ms step_avg:204.05ms
step:3139/4740 train_time:640477ms step_avg:204.04ms
step:3140/4740 train_time:640798ms step_avg:204.08ms
step:3141/4740 train_time:640976ms step_avg:204.07ms
step:3142/4740 train_time:641299ms step_avg:204.11ms
step:3143/4740 train_time:641476ms step_avg:204.10ms
step:3144/4740 train_time:641796ms step_avg:204.13ms
step:3145/4740 train_time:641978ms step_avg:204.13ms
step:3146/4740 train_time:642300ms step_avg:204.16ms
step:3147/4740 train_time:642478ms step_avg:204.16ms
step:3148/4740 train_time:642799ms step_avg:204.19ms
step:3149/4740 train_time:642978ms step_avg:204.18ms
step:3150/4740 train_time:643298ms step_avg:204.22ms
step:3151/4740 train_time:643476ms step_avg:204.21ms
step:3152/4740 train_time:643798ms step_avg:204.25ms
step:3153/4740 train_time:643978ms step_avg:204.24ms
step:3154/4740 train_time:644299ms step_avg:204.28ms
step:3155/4740 train_time:644478ms step_avg:204.27ms
step:3156/4740 train_time:644798ms step_avg:204.31ms
step:3157/4740 train_time:644976ms step_avg:204.30ms
step:3158/4740 train_time:645297ms step_avg:204.34ms
step:3159/4740 train_time:645476ms step_avg:204.33ms
step:3160/4740 train_time:645796ms step_avg:204.37ms
step:3161/4740 train_time:645975ms step_avg:204.36ms
step:3162/4740 train_time:646299ms step_avg:204.40ms
step:3163/4740 train_time:646475ms step_avg:204.39ms
step:3164/4740 train_time:646798ms step_avg:204.42ms
step:3165/4740 train_time:646975ms step_avg:204.42ms
step:3166/4740 train_time:647297ms step_avg:204.45ms
step:3167/4740 train_time:647475ms step_avg:204.44ms
step:3168/4740 train_time:647798ms step_avg:204.48ms
step:3169/4740 train_time:647975ms step_avg:204.47ms
step:3170/4740 train_time:648294ms step_avg:204.51ms
step:3171/4740 train_time:648474ms step_avg:204.50ms
step:3172/4740 train_time:648795ms step_avg:204.54ms
step:3173/4740 train_time:648974ms step_avg:204.53ms
step:3174/4740 train_time:649296ms step_avg:204.57ms
step:3175/4740 train_time:649473ms step_avg:204.56ms
step:3176/4740 train_time:649797ms step_avg:204.60ms
step:3177/4740 train_time:649976ms step_avg:204.59ms
step:3178/4740 train_time:650299ms step_avg:204.63ms
step:3179/4740 train_time:650476ms step_avg:204.62ms
step:3180/4740 train_time:650799ms step_avg:204.65ms
step:3181/4740 train_time:650978ms step_avg:204.65ms
step:3182/4740 train_time:651301ms step_avg:204.68ms
step:3183/4740 train_time:651480ms step_avg:204.67ms
step:3184/4740 train_time:651803ms step_avg:204.71ms
step:3185/4740 train_time:651979ms step_avg:204.70ms
step:3186/4740 train_time:652299ms step_avg:204.74ms
step:3187/4740 train_time:652478ms step_avg:204.73ms
step:3188/4740 train_time:652800ms step_avg:204.77ms
step:3189/4740 train_time:652981ms step_avg:204.76ms
step:3190/4740 train_time:653303ms step_avg:204.80ms
step:3191/4740 train_time:653484ms step_avg:204.79ms
step:3192/4740 train_time:653802ms step_avg:204.83ms
step:3193/4740 train_time:653980ms step_avg:204.82ms
step:3194/4740 train_time:654303ms step_avg:204.85ms
step:3195/4740 train_time:654480ms step_avg:204.85ms
step:3196/4740 train_time:654804ms step_avg:204.88ms
step:3197/4740 train_time:654982ms step_avg:204.87ms
step:3198/4740 train_time:655305ms step_avg:204.91ms
step:3199/4740 train_time:655483ms step_avg:204.90ms
step:3200/4740 train_time:655803ms step_avg:204.94ms
step:3201/4740 train_time:655979ms step_avg:204.93ms
step:3202/4740 train_time:656296ms step_avg:204.96ms
step:3203/4740 train_time:656478ms step_avg:204.96ms
step:3204/4740 train_time:656798ms step_avg:204.99ms
step:3205/4740 train_time:656978ms step_avg:204.99ms
step:3206/4740 train_time:657299ms step_avg:205.02ms
step:3207/4740 train_time:657481ms step_avg:205.01ms
step:3208/4740 train_time:657803ms step_avg:205.05ms
step:3209/4740 train_time:657979ms step_avg:205.04ms
step:3210/4740 train_time:658300ms step_avg:205.08ms
step:3211/4740 train_time:658481ms step_avg:205.07ms
step:3212/4740 train_time:658801ms step_avg:205.11ms
step:3213/4740 train_time:658977ms step_avg:205.10ms
step:3214/4740 train_time:659300ms step_avg:205.13ms
step:3215/4740 train_time:659480ms step_avg:205.13ms
step:3216/4740 train_time:659803ms step_avg:205.16ms
step:3217/4740 train_time:659980ms step_avg:205.15ms
step:3218/4740 train_time:660299ms step_avg:205.19ms
step:3219/4740 train_time:660480ms step_avg:205.18ms
step:3220/4740 train_time:660805ms step_avg:205.22ms
step:3221/4740 train_time:660983ms step_avg:205.21ms
step:3222/4740 train_time:661305ms step_avg:205.25ms
step:3223/4740 train_time:661482ms step_avg:205.24ms
step:3224/4740 train_time:661804ms step_avg:205.27ms
step:3225/4740 train_time:661985ms step_avg:205.27ms
step:3226/4740 train_time:662308ms step_avg:205.30ms
step:3227/4740 train_time:662486ms step_avg:205.29ms
step:3228/4740 train_time:662810ms step_avg:205.33ms
step:3229/4740 train_time:662987ms step_avg:205.32ms
step:3230/4740 train_time:663310ms step_avg:205.36ms
step:3231/4740 train_time:663489ms step_avg:205.35ms
step:3232/4740 train_time:663807ms step_avg:205.39ms
step:3233/4740 train_time:663988ms step_avg:205.38ms
step:3234/4740 train_time:664310ms step_avg:205.41ms
step:3235/4740 train_time:664488ms step_avg:205.41ms
step:3236/4740 train_time:664807ms step_avg:205.44ms
step:3237/4740 train_time:664987ms step_avg:205.43ms
step:3238/4740 train_time:665304ms step_avg:205.47ms
step:3239/4740 train_time:665484ms step_avg:205.46ms
step:3240/4740 train_time:665804ms step_avg:205.50ms
step:3241/4740 train_time:665986ms step_avg:205.49ms
step:3242/4740 train_time:666307ms step_avg:205.52ms
step:3243/4740 train_time:666488ms step_avg:205.52ms
step:3244/4740 train_time:666810ms step_avg:205.55ms
step:3245/4740 train_time:666989ms step_avg:205.54ms
step:3246/4740 train_time:667308ms step_avg:205.58ms
step:3247/4740 train_time:667487ms step_avg:205.57ms
step:3248/4740 train_time:667806ms step_avg:205.61ms
step:3249/4740 train_time:667986ms step_avg:205.60ms
step:3250/4740 train_time:668306ms step_avg:205.63ms
step:3250/4740 val_loss:3.0678 train_time:668432ms step_avg:205.67ms
step:3251/4740 train_time:668488ms step_avg:205.63ms
step:3252/4740 train_time:668814ms step_avg:205.66ms
step:3253/4740 train_time:668993ms step_avg:205.65ms
step:3254/4740 train_time:669309ms step_avg:205.69ms
step:3255/4740 train_time:669487ms step_avg:205.68ms
step:3256/4740 train_time:669811ms step_avg:205.72ms
step:3257/4740 train_time:669996ms step_avg:205.71ms
step:3258/4740 train_time:670313ms step_avg:205.74ms
step:3259/4740 train_time:670492ms step_avg:205.74ms
step:3260/4740 train_time:670813ms step_avg:205.77ms
step:3261/4740 train_time:670996ms step_avg:205.76ms
step:3262/4740 train_time:671314ms step_avg:205.80ms
step:3263/4740 train_time:671496ms step_avg:205.79ms
step:3264/4740 train_time:671814ms step_avg:205.83ms
step:3265/4740 train_time:671994ms step_avg:205.82ms
step:3266/4740 train_time:672313ms step_avg:205.85ms
step:3267/4740 train_time:672493ms step_avg:205.84ms
step:3268/4740 train_time:672813ms step_avg:205.88ms
step:3269/4740 train_time:672993ms step_avg:205.87ms
step:3270/4740 train_time:673314ms step_avg:205.91ms
step:3271/4740 train_time:673492ms step_avg:205.90ms
step:3272/4740 train_time:673811ms step_avg:205.93ms
step:3273/4740 train_time:673991ms step_avg:205.92ms
step:3274/4740 train_time:674311ms step_avg:205.96ms
step:3275/4740 train_time:674491ms step_avg:205.95ms
step:3276/4740 train_time:674812ms step_avg:205.99ms
step:3277/4740 train_time:674994ms step_avg:205.98ms
step:3278/4740 train_time:675311ms step_avg:206.01ms
step:3279/4740 train_time:675492ms step_avg:206.01ms
step:3280/4740 train_time:675814ms step_avg:206.04ms
step:3281/4740 train_time:675994ms step_avg:206.03ms
step:3282/4740 train_time:676313ms step_avg:206.07ms
step:3283/4740 train_time:676493ms step_avg:206.06ms
step:3284/4740 train_time:676816ms step_avg:206.09ms
step:3285/4740 train_time:676995ms step_avg:206.09ms
step:3286/4740 train_time:677311ms step_avg:206.12ms
step:3287/4740 train_time:677492ms step_avg:206.11ms
step:3288/4740 train_time:677813ms step_avg:206.15ms
step:3289/4740 train_time:677995ms step_avg:206.14ms
step:3290/4740 train_time:678316ms step_avg:206.17ms
step:3291/4740 train_time:678497ms step_avg:206.17ms
step:3292/4740 train_time:678814ms step_avg:206.20ms
step:3293/4740 train_time:678994ms step_avg:206.19ms
step:3294/4740 train_time:679313ms step_avg:206.23ms
step:3295/4740 train_time:679490ms step_avg:206.22ms
step:3296/4740 train_time:679812ms step_avg:206.25ms
step:3297/4740 train_time:679992ms step_avg:206.25ms
step:3298/4740 train_time:680312ms step_avg:206.28ms
step:3299/4740 train_time:680489ms step_avg:206.27ms
step:3300/4740 train_time:680811ms step_avg:206.31ms
step:3301/4740 train_time:680989ms step_avg:206.30ms
step:3302/4740 train_time:681310ms step_avg:206.33ms
step:3303/4740 train_time:681486ms step_avg:206.32ms
step:3304/4740 train_time:681808ms step_avg:206.36ms
step:3305/4740 train_time:681989ms step_avg:206.35ms
step:3306/4740 train_time:682309ms step_avg:206.39ms
step:3307/4740 train_time:682493ms step_avg:206.38ms
step:3308/4740 train_time:682813ms step_avg:206.41ms
step:3309/4740 train_time:682994ms step_avg:206.40ms
step:3310/4740 train_time:683312ms step_avg:206.44ms
step:3311/4740 train_time:683492ms step_avg:206.43ms
step:3312/4740 train_time:683813ms step_avg:206.47ms
step:3313/4740 train_time:683992ms step_avg:206.46ms
step:3314/4740 train_time:684310ms step_avg:206.49ms
step:3315/4740 train_time:684495ms step_avg:206.48ms
step:3316/4740 train_time:684816ms step_avg:206.52ms
step:3317/4740 train_time:684997ms step_avg:206.51ms
step:3318/4740 train_time:685317ms step_avg:206.55ms
step:3319/4740 train_time:685497ms step_avg:206.54ms
step:3320/4740 train_time:685818ms step_avg:206.57ms
step:3321/4740 train_time:685996ms step_avg:206.56ms
step:3322/4740 train_time:686317ms step_avg:206.60ms
step:3323/4740 train_time:686496ms step_avg:206.59ms
step:3324/4740 train_time:686817ms step_avg:206.62ms
step:3325/4740 train_time:686995ms step_avg:206.62ms
step:3326/4740 train_time:687313ms step_avg:206.65ms
step:3327/4740 train_time:687493ms step_avg:206.64ms
step:3328/4740 train_time:687814ms step_avg:206.67ms
step:3329/4740 train_time:687991ms step_avg:206.67ms
step:3330/4740 train_time:688312ms step_avg:206.70ms
step:3331/4740 train_time:688487ms step_avg:206.69ms
step:3332/4740 train_time:688811ms step_avg:206.73ms
step:3333/4740 train_time:688990ms step_avg:206.72ms
step:3334/4740 train_time:689311ms step_avg:206.75ms
step:3335/4740 train_time:689488ms step_avg:206.74ms
step:3336/4740 train_time:689812ms step_avg:206.78ms
step:3337/4740 train_time:689988ms step_avg:206.77ms
step:3338/4740 train_time:690309ms step_avg:206.80ms
step:3339/4740 train_time:690487ms step_avg:206.79ms
step:3340/4740 train_time:690807ms step_avg:206.83ms
step:3341/4740 train_time:690987ms step_avg:206.82ms
step:3342/4740 train_time:691310ms step_avg:206.86ms
step:3343/4740 train_time:691488ms step_avg:206.85ms
step:3344/4740 train_time:691808ms step_avg:206.88ms
step:3345/4740 train_time:691986ms step_avg:206.87ms
step:3346/4740 train_time:692309ms step_avg:206.91ms
step:3347/4740 train_time:692490ms step_avg:206.90ms
step:3348/4740 train_time:692811ms step_avg:206.93ms
step:3349/4740 train_time:692989ms step_avg:206.92ms
step:3350/4740 train_time:693313ms step_avg:206.96ms
step:3351/4740 train_time:693495ms step_avg:206.95ms
step:3352/4740 train_time:693817ms step_avg:206.99ms
step:3353/4740 train_time:693997ms step_avg:206.98ms
step:3354/4740 train_time:694315ms step_avg:207.01ms
step:3355/4740 train_time:694495ms step_avg:207.00ms
step:3356/4740 train_time:694813ms step_avg:207.04ms
step:3357/4740 train_time:694993ms step_avg:207.03ms
step:3358/4740 train_time:695316ms step_avg:207.06ms
step:3359/4740 train_time:695494ms step_avg:207.05ms
step:3360/4740 train_time:695815ms step_avg:207.09ms
step:3361/4740 train_time:695992ms step_avg:207.08ms
step:3362/4740 train_time:696314ms step_avg:207.11ms
step:3363/4740 train_time:696492ms step_avg:207.10ms
step:3364/4740 train_time:696817ms step_avg:207.14ms
step:3365/4740 train_time:696997ms step_avg:207.13ms
step:3366/4740 train_time:697317ms step_avg:207.16ms
step:3367/4740 train_time:697493ms step_avg:207.16ms
step:3368/4740 train_time:697814ms step_avg:207.19ms
step:3369/4740 train_time:697995ms step_avg:207.18ms
step:3370/4740 train_time:698316ms step_avg:207.22ms
step:3371/4740 train_time:698496ms step_avg:207.21ms
step:3372/4740 train_time:698817ms step_avg:207.24ms
step:3373/4740 train_time:698998ms step_avg:207.23ms
step:3374/4740 train_time:699318ms step_avg:207.27ms
step:3375/4740 train_time:699496ms step_avg:207.26ms
step:3376/4740 train_time:699814ms step_avg:207.29ms
step:3377/4740 train_time:699994ms step_avg:207.28ms
step:3378/4740 train_time:700313ms step_avg:207.32ms
step:3379/4740 train_time:700492ms step_avg:207.31ms
step:3380/4740 train_time:700812ms step_avg:207.34ms
step:3381/4740 train_time:700994ms step_avg:207.33ms
step:3382/4740 train_time:701312ms step_avg:207.37ms
step:3383/4740 train_time:701493ms step_avg:207.36ms
step:3384/4740 train_time:701814ms step_avg:207.39ms
step:3385/4740 train_time:701993ms step_avg:207.38ms
step:3386/4740 train_time:702315ms step_avg:207.42ms
step:3387/4740 train_time:702495ms step_avg:207.41ms
step:3388/4740 train_time:702817ms step_avg:207.44ms
step:3389/4740 train_time:703001ms step_avg:207.44ms
step:3390/4740 train_time:703321ms step_avg:207.47ms
step:3391/4740 train_time:703500ms step_avg:207.46ms
step:3392/4740 train_time:703820ms step_avg:207.49ms
step:3393/4740 train_time:704002ms step_avg:207.49ms
step:3394/4740 train_time:704322ms step_avg:207.52ms
step:3395/4740 train_time:704503ms step_avg:207.51ms
step:3396/4740 train_time:704826ms step_avg:207.55ms
step:3397/4740 train_time:705007ms step_avg:207.54ms
step:3398/4740 train_time:705331ms step_avg:207.57ms
step:3399/4740 train_time:705508ms step_avg:207.56ms
step:3400/4740 train_time:705831ms step_avg:207.60ms
step:3401/4740 train_time:706010ms step_avg:207.59ms
step:3402/4740 train_time:706335ms step_avg:207.62ms
step:3403/4740 train_time:706512ms step_avg:207.61ms
step:3404/4740 train_time:706835ms step_avg:207.65ms
step:3405/4740 train_time:707010ms step_avg:207.64ms
step:3406/4740 train_time:707333ms step_avg:207.67ms
step:3407/4740 train_time:707508ms step_avg:207.66ms
step:3408/4740 train_time:707829ms step_avg:207.70ms
step:3409/4740 train_time:708011ms step_avg:207.69ms
step:3410/4740 train_time:708332ms step_avg:207.72ms
step:3411/4740 train_time:708512ms step_avg:207.71ms
step:3412/4740 train_time:708834ms step_avg:207.75ms
step:3413/4740 train_time:709017ms step_avg:207.74ms
step:3414/4740 train_time:709334ms step_avg:207.77ms
step:3415/4740 train_time:709515ms step_avg:207.76ms
step:3416/4740 train_time:709839ms step_avg:207.80ms
step:3417/4740 train_time:710020ms step_avg:207.79ms
step:3418/4740 train_time:710338ms step_avg:207.82ms
step:3419/4740 train_time:710515ms step_avg:207.81ms
step:3420/4740 train_time:710838ms step_avg:207.85ms
step:3421/4740 train_time:711015ms step_avg:207.84ms
step:3422/4740 train_time:711339ms step_avg:207.87ms
step:3423/4740 train_time:711514ms step_avg:207.86ms
step:3424/4740 train_time:711836ms step_avg:207.90ms
step:3425/4740 train_time:712011ms step_avg:207.89ms
step:3426/4740 train_time:712335ms step_avg:207.92ms
step:3427/4740 train_time:712511ms step_avg:207.91ms
step:3428/4740 train_time:712834ms step_avg:207.94ms
step:3429/4740 train_time:713013ms step_avg:207.94ms
step:3430/4740 train_time:713334ms step_avg:207.97ms
step:3431/4740 train_time:713513ms step_avg:207.96ms
step:3432/4740 train_time:713837ms step_avg:207.99ms
step:3433/4740 train_time:714013ms step_avg:207.99ms
step:3434/4740 train_time:714336ms step_avg:208.02ms
step:3435/4740 train_time:714512ms step_avg:208.01ms
step:3436/4740 train_time:714834ms step_avg:208.04ms
step:3437/4740 train_time:715009ms step_avg:208.03ms
step:3438/4740 train_time:715332ms step_avg:208.07ms
step:3439/4740 train_time:715510ms step_avg:208.06ms
step:3440/4740 train_time:715834ms step_avg:208.09ms
step:3441/4740 train_time:716012ms step_avg:208.08ms
step:3442/4740 train_time:716335ms step_avg:208.12ms
step:3443/4740 train_time:716511ms step_avg:208.11ms
step:3444/4740 train_time:716834ms step_avg:208.14ms
step:3445/4740 train_time:717013ms step_avg:208.13ms
step:3446/4740 train_time:717337ms step_avg:208.17ms
step:3447/4740 train_time:717514ms step_avg:208.16ms
step:3448/4740 train_time:717839ms step_avg:208.19ms
step:3449/4740 train_time:718019ms step_avg:208.18ms
step:3450/4740 train_time:718339ms step_avg:208.21ms
step:3451/4740 train_time:718517ms step_avg:208.21ms
step:3452/4740 train_time:718839ms step_avg:208.24ms
step:3453/4740 train_time:719016ms step_avg:208.23ms
step:3454/4740 train_time:719335ms step_avg:208.26ms
step:3455/4740 train_time:719515ms step_avg:208.25ms
step:3456/4740 train_time:719841ms step_avg:208.29ms
step:3457/4740 train_time:720015ms step_avg:208.28ms
step:3458/4740 train_time:720337ms step_avg:208.31ms
step:3459/4740 train_time:720517ms step_avg:208.30ms
step:3460/4740 train_time:720839ms step_avg:208.33ms
step:3461/4740 train_time:721017ms step_avg:208.33ms
step:3462/4740 train_time:721341ms step_avg:208.36ms
step:3463/4740 train_time:721517ms step_avg:208.35ms
step:3464/4740 train_time:721841ms step_avg:208.38ms
step:3465/4740 train_time:722016ms step_avg:208.37ms
step:3466/4740 train_time:722342ms step_avg:208.41ms
step:3467/4740 train_time:722517ms step_avg:208.40ms
step:3468/4740 train_time:722836ms step_avg:208.43ms
step:3469/4740 train_time:723017ms step_avg:208.42ms
step:3470/4740 train_time:723340ms step_avg:208.46ms
step:3471/4740 train_time:723516ms step_avg:208.45ms
step:3472/4740 train_time:723837ms step_avg:208.48ms
step:3473/4740 train_time:724016ms step_avg:208.47ms
step:3474/4740 train_time:724339ms step_avg:208.50ms
step:3475/4740 train_time:724513ms step_avg:208.49ms
step:3476/4740 train_time:724833ms step_avg:208.53ms
step:3477/4740 train_time:725013ms step_avg:208.52ms
step:3478/4740 train_time:725336ms step_avg:208.55ms
step:3479/4740 train_time:725513ms step_avg:208.54ms
step:3480/4740 train_time:725836ms step_avg:208.57ms
step:3481/4740 train_time:726014ms step_avg:208.56ms
step:3482/4740 train_time:726340ms step_avg:208.60ms
step:3483/4740 train_time:726515ms step_avg:208.59ms
step:3484/4740 train_time:726839ms step_avg:208.62ms
step:3485/4740 train_time:727015ms step_avg:208.61ms
step:3486/4740 train_time:727339ms step_avg:208.65ms
step:3487/4740 train_time:727519ms step_avg:208.64ms
step:3488/4740 train_time:727841ms step_avg:208.67ms
step:3489/4740 train_time:728020ms step_avg:208.66ms
step:3490/4740 train_time:728339ms step_avg:208.69ms
step:3491/4740 train_time:728517ms step_avg:208.68ms
step:3492/4740 train_time:728839ms step_avg:208.72ms
step:3493/4740 train_time:729014ms step_avg:208.71ms
step:3494/4740 train_time:729338ms step_avg:208.74ms
step:3495/4740 train_time:729513ms step_avg:208.73ms
step:3496/4740 train_time:729834ms step_avg:208.76ms
step:3497/4740 train_time:730013ms step_avg:208.75ms
step:3498/4740 train_time:730337ms step_avg:208.79ms
step:3499/4740 train_time:730516ms step_avg:208.78ms
step:3500/4740 train_time:730842ms step_avg:208.81ms
step:3500/4740 val_loss:3.0361 train_time:730965ms step_avg:208.85ms
step:3501/4740 train_time:731026ms step_avg:208.80ms
step:3502/4740 train_time:731360ms step_avg:208.84ms
step:3503/4740 train_time:731535ms step_avg:208.83ms
step:3504/4740 train_time:731848ms step_avg:208.86ms
step:3505/4740 train_time:732021ms step_avg:208.85ms
step:3506/4740 train_time:732351ms step_avg:208.88ms
step:3507/4740 train_time:732531ms step_avg:208.88ms
step:3508/4740 train_time:732850ms step_avg:208.91ms
step:3509/4740 train_time:733026ms step_avg:208.90ms
step:3510/4740 train_time:733352ms step_avg:208.93ms
step:3511/4740 train_time:733531ms step_avg:208.92ms
step:3512/4740 train_time:733849ms step_avg:208.95ms
step:3513/4740 train_time:734025ms step_avg:208.95ms
step:3514/4740 train_time:734348ms step_avg:208.98ms
step:3515/4740 train_time:734526ms step_avg:208.97ms
step:3516/4740 train_time:734843ms step_avg:209.00ms
step:3517/4740 train_time:735023ms step_avg:208.99ms
step:3518/4740 train_time:735345ms step_avg:209.02ms
step:3519/4740 train_time:735523ms step_avg:209.01ms
step:3520/4740 train_time:735845ms step_avg:209.05ms
step:3521/4740 train_time:736026ms step_avg:209.04ms
step:3522/4740 train_time:736346ms step_avg:209.07ms
step:3523/4740 train_time:736522ms step_avg:209.06ms
step:3524/4740 train_time:736846ms step_avg:209.09ms
step:3525/4740 train_time:737023ms step_avg:209.08ms
step:3526/4740 train_time:737346ms step_avg:209.12ms
step:3527/4740 train_time:737524ms step_avg:209.11ms
step:3528/4740 train_time:737844ms step_avg:209.14ms
step:3529/4740 train_time:738024ms step_avg:209.13ms
step:3530/4740 train_time:738346ms step_avg:209.16ms
step:3531/4740 train_time:738522ms step_avg:209.15ms
step:3532/4740 train_time:738846ms step_avg:209.19ms
step:3533/4740 train_time:739023ms step_avg:209.18ms
step:3534/4740 train_time:739342ms step_avg:209.21ms
step:3535/4740 train_time:739520ms step_avg:209.20ms
step:3536/4740 train_time:739844ms step_avg:209.23ms
step:3537/4740 train_time:740020ms step_avg:209.22ms
step:3538/4740 train_time:740343ms step_avg:209.25ms
step:3539/4740 train_time:740520ms step_avg:209.25ms
step:3540/4740 train_time:740844ms step_avg:209.28ms
step:3541/4740 train_time:741026ms step_avg:209.27ms
step:3542/4740 train_time:741349ms step_avg:209.30ms
step:3543/4740 train_time:741524ms step_avg:209.29ms
step:3544/4740 train_time:741844ms step_avg:209.32ms
step:3545/4740 train_time:742024ms step_avg:209.32ms
step:3546/4740 train_time:742344ms step_avg:209.35ms
step:3547/4740 train_time:742528ms step_avg:209.34ms
step:3548/4740 train_time:742850ms step_avg:209.37ms
step:3549/4740 train_time:743028ms step_avg:209.36ms
step:3550/4740 train_time:743351ms step_avg:209.39ms
step:3551/4740 train_time:743531ms step_avg:209.39ms
step:3552/4740 train_time:743848ms step_avg:209.42ms
step:3553/4740 train_time:744026ms step_avg:209.41ms
step:3554/4740 train_time:744344ms step_avg:209.44ms
step:3555/4740 train_time:744525ms step_avg:209.43ms
step:3556/4740 train_time:744853ms step_avg:209.46ms
step:3557/4740 train_time:745030ms step_avg:209.45ms
step:3558/4740 train_time:745349ms step_avg:209.49ms
step:3559/4740 train_time:745530ms step_avg:209.48ms
step:3560/4740 train_time:745851ms step_avg:209.51ms
step:3561/4740 train_time:746030ms step_avg:209.50ms
step:3562/4740 train_time:746349ms step_avg:209.53ms
step:3563/4740 train_time:746525ms step_avg:209.52ms
step:3564/4740 train_time:746848ms step_avg:209.55ms
step:3565/4740 train_time:747025ms step_avg:209.54ms
step:3566/4740 train_time:747346ms step_avg:209.58ms
step:3567/4740 train_time:747523ms step_avg:209.57ms
step:3568/4740 train_time:747847ms step_avg:209.60ms
step:3569/4740 train_time:748021ms step_avg:209.59ms
step:3570/4740 train_time:748343ms step_avg:209.62ms
step:3571/4740 train_time:748518ms step_avg:209.61ms
step:3572/4740 train_time:748845ms step_avg:209.64ms
step:3573/4740 train_time:749021ms step_avg:209.63ms
step:3574/4740 train_time:749337ms step_avg:209.66ms
step:3575/4740 train_time:749521ms step_avg:209.66ms
step:3576/4740 train_time:749844ms step_avg:209.69ms
step:3577/4740 train_time:750021ms step_avg:209.68ms
step:3578/4740 train_time:750343ms step_avg:209.71ms
step:3579/4740 train_time:750523ms step_avg:209.70ms
step:3580/4740 train_time:750846ms step_avg:209.73ms
step:3581/4740 train_time:751024ms step_avg:209.72ms
step:3582/4740 train_time:751345ms step_avg:209.76ms
step:3583/4740 train_time:751530ms step_avg:209.75ms
step:3584/4740 train_time:751852ms step_avg:209.78ms
step:3585/4740 train_time:752028ms step_avg:209.77ms
step:3586/4740 train_time:752350ms step_avg:209.80ms
step:3587/4740 train_time:752529ms step_avg:209.79ms
step:3588/4740 train_time:752847ms step_avg:209.82ms
step:3589/4740 train_time:753025ms step_avg:209.81ms
step:3590/4740 train_time:753346ms step_avg:209.85ms
step:3591/4740 train_time:753527ms step_avg:209.84ms
step:3592/4740 train_time:753849ms step_avg:209.87ms
step:3593/4740 train_time:754030ms step_avg:209.86ms
step:3594/4740 train_time:754348ms step_avg:209.89ms
step:3595/4740 train_time:754526ms step_avg:209.88ms
step:3596/4740 train_time:754845ms step_avg:209.91ms
step:3597/4740 train_time:755024ms step_avg:209.90ms
step:3598/4740 train_time:755346ms step_avg:209.94ms
step:3599/4740 train_time:755524ms step_avg:209.93ms
step:3600/4740 train_time:755844ms step_avg:209.96ms
step:3601/4740 train_time:756020ms step_avg:209.95ms
step:3602/4740 train_time:756342ms step_avg:209.98ms
step:3603/4740 train_time:756520ms step_avg:209.97ms
step:3604/4740 train_time:756843ms step_avg:210.00ms
step:3605/4740 train_time:757019ms step_avg:209.99ms
step:3606/4740 train_time:757339ms step_avg:210.02ms
step:3607/4740 train_time:757517ms step_avg:210.01ms
step:3608/4740 train_time:757839ms step_avg:210.04ms
step:3609/4740 train_time:758017ms step_avg:210.04ms
step:3610/4740 train_time:758337ms step_avg:210.07ms
step:3611/4740 train_time:758516ms step_avg:210.06ms
step:3612/4740 train_time:758838ms step_avg:210.09ms
step:3613/4740 train_time:759016ms step_avg:210.08ms
step:3614/4740 train_time:759339ms step_avg:210.11ms
step:3615/4740 train_time:759515ms step_avg:210.10ms
step:3616/4740 train_time:759837ms step_avg:210.13ms
step:3617/4740 train_time:760019ms step_avg:210.12ms
step:3618/4740 train_time:760343ms step_avg:210.16ms
step:3619/4740 train_time:760519ms step_avg:210.15ms
step:3620/4740 train_time:760839ms step_avg:210.18ms
step:3621/4740 train_time:761014ms step_avg:210.17ms
step:3622/4740 train_time:761337ms step_avg:210.20ms
step:3623/4740 train_time:761517ms step_avg:210.19ms
step:3624/4740 train_time:761838ms step_avg:210.22ms
step:3625/4740 train_time:762020ms step_avg:210.21ms
step:3626/4740 train_time:762338ms step_avg:210.24ms
step:3627/4740 train_time:762519ms step_avg:210.23ms
step:3628/4740 train_time:762840ms step_avg:210.26ms
step:3629/4740 train_time:763022ms step_avg:210.26ms
step:3630/4740 train_time:763342ms step_avg:210.29ms
step:3631/4740 train_time:763519ms step_avg:210.28ms
step:3632/4740 train_time:763838ms step_avg:210.31ms
step:3633/4740 train_time:764018ms step_avg:210.30ms
step:3634/4740 train_time:764342ms step_avg:210.33ms
step:3635/4740 train_time:764520ms step_avg:210.32ms
step:3636/4740 train_time:764841ms step_avg:210.35ms
step:3637/4740 train_time:765022ms step_avg:210.34ms
step:3638/4740 train_time:765343ms step_avg:210.37ms
step:3639/4740 train_time:765519ms step_avg:210.37ms
step:3640/4740 train_time:765840ms step_avg:210.40ms
step:3641/4740 train_time:766019ms step_avg:210.39ms
step:3642/4740 train_time:766340ms step_avg:210.42ms
step:3643/4740 train_time:766520ms step_avg:210.41ms
step:3644/4740 train_time:766838ms step_avg:210.44ms
step:3645/4740 train_time:767020ms step_avg:210.43ms
step:3646/4740 train_time:767340ms step_avg:210.46ms
step:3647/4740 train_time:767520ms step_avg:210.45ms
step:3648/4740 train_time:767840ms step_avg:210.48ms
step:3649/4740 train_time:768020ms step_avg:210.47ms
step:3650/4740 train_time:768344ms step_avg:210.51ms
step:3651/4740 train_time:768524ms step_avg:210.50ms
step:3652/4740 train_time:768846ms step_avg:210.53ms
step:3653/4740 train_time:769023ms step_avg:210.52ms
step:3654/4740 train_time:769346ms step_avg:210.55ms
step:3655/4740 train_time:769525ms step_avg:210.54ms
step:3656/4740 train_time:769845ms step_avg:210.57ms
step:3657/4740 train_time:770024ms step_avg:210.56ms
step:3658/4740 train_time:770343ms step_avg:210.59ms
step:3659/4740 train_time:770523ms step_avg:210.58ms
step:3660/4740 train_time:770843ms step_avg:210.61ms
step:3661/4740 train_time:771027ms step_avg:210.61ms
step:3662/4740 train_time:771347ms step_avg:210.64ms
step:3663/4740 train_time:771524ms step_avg:210.63ms
step:3664/4740 train_time:771846ms step_avg:210.66ms
step:3665/4740 train_time:772026ms step_avg:210.65ms
step:3666/4740 train_time:772348ms step_avg:210.68ms
step:3667/4740 train_time:772528ms step_avg:210.67ms
step:3668/4740 train_time:772847ms step_avg:210.70ms
step:3669/4740 train_time:773023ms step_avg:210.69ms
step:3670/4740 train_time:773344ms step_avg:210.72ms
step:3671/4740 train_time:773523ms step_avg:210.71ms
step:3672/4740 train_time:773846ms step_avg:210.74ms
step:3673/4740 train_time:774024ms step_avg:210.73ms
step:3674/4740 train_time:774346ms step_avg:210.76ms
step:3675/4740 train_time:774521ms step_avg:210.75ms
step:3676/4740 train_time:774847ms step_avg:210.79ms
step:3677/4740 train_time:775022ms step_avg:210.78ms
step:3678/4740 train_time:775349ms step_avg:210.81ms
step:3679/4740 train_time:775525ms step_avg:210.80ms
step:3680/4740 train_time:775849ms step_avg:210.83ms
step:3681/4740 train_time:776027ms step_avg:210.82ms
step:3682/4740 train_time:776348ms step_avg:210.85ms
step:3683/4740 train_time:776525ms step_avg:210.84ms
step:3684/4740 train_time:776844ms step_avg:210.87ms
step:3685/4740 train_time:777024ms step_avg:210.86ms
step:3686/4740 train_time:777341ms step_avg:210.89ms
step:3687/4740 train_time:777524ms step_avg:210.88ms
step:3688/4740 train_time:777844ms step_avg:210.91ms
step:3689/4740 train_time:778024ms step_avg:210.90ms
step:3690/4740 train_time:778347ms step_avg:210.93ms
step:3691/4740 train_time:778522ms step_avg:210.92ms
step:3692/4740 train_time:778844ms step_avg:210.95ms
step:3693/4740 train_time:779024ms step_avg:210.95ms
step:3694/4740 train_time:779351ms step_avg:210.98ms
step:3695/4740 train_time:779527ms step_avg:210.97ms
step:3696/4740 train_time:779848ms step_avg:211.00ms
step:3697/4740 train_time:780026ms step_avg:210.99ms
step:3698/4740 train_time:780346ms step_avg:211.02ms
step:3699/4740 train_time:780523ms step_avg:211.01ms
step:3700/4740 train_time:780844ms step_avg:211.04ms
step:3701/4740 train_time:781020ms step_avg:211.03ms
step:3702/4740 train_time:781343ms step_avg:211.06ms
step:3703/4740 train_time:781524ms step_avg:211.05ms
step:3704/4740 train_time:781844ms step_avg:211.08ms
step:3705/4740 train_time:782022ms step_avg:211.07ms
step:3706/4740 train_time:782345ms step_avg:211.10ms
step:3707/4740 train_time:782521ms step_avg:211.09ms
step:3708/4740 train_time:782843ms step_avg:211.12ms
step:3709/4740 train_time:783020ms step_avg:211.11ms
step:3710/4740 train_time:783343ms step_avg:211.14ms
step:3711/4740 train_time:783521ms step_avg:211.13ms
step:3712/4740 train_time:783845ms step_avg:211.17ms
step:3713/4740 train_time:784021ms step_avg:211.16ms
step:3714/4740 train_time:784347ms step_avg:211.19ms
step:3715/4740 train_time:784523ms step_avg:211.18ms
step:3716/4740 train_time:784847ms step_avg:211.21ms
step:3717/4740 train_time:785025ms step_avg:211.20ms
step:3718/4740 train_time:785346ms step_avg:211.23ms
step:3719/4740 train_time:785521ms step_avg:211.22ms
step:3720/4740 train_time:785843ms step_avg:211.25ms
step:3721/4740 train_time:786019ms step_avg:211.24ms
step:3722/4740 train_time:786343ms step_avg:211.27ms
step:3723/4740 train_time:786519ms step_avg:211.26ms
step:3724/4740 train_time:786844ms step_avg:211.29ms
step:3725/4740 train_time:787022ms step_avg:211.28ms
step:3726/4740 train_time:787345ms step_avg:211.31ms
step:3727/4740 train_time:787524ms step_avg:211.30ms
step:3728/4740 train_time:787845ms step_avg:211.33ms
step:3729/4740 train_time:788020ms step_avg:211.32ms
step:3730/4740 train_time:788342ms step_avg:211.35ms
step:3731/4740 train_time:788521ms step_avg:211.34ms
step:3732/4740 train_time:788841ms step_avg:211.37ms
step:3733/4740 train_time:789019ms step_avg:211.36ms
step:3734/4740 train_time:789344ms step_avg:211.39ms
step:3735/4740 train_time:789521ms step_avg:211.38ms
step:3736/4740 train_time:789842ms step_avg:211.41ms
step:3737/4740 train_time:790019ms step_avg:211.40ms
step:3738/4740 train_time:790340ms step_avg:211.43ms
step:3739/4740 train_time:790519ms step_avg:211.43ms
step:3740/4740 train_time:790841ms step_avg:211.45ms
step:3741/4740 train_time:791016ms step_avg:211.45ms
step:3742/4740 train_time:791337ms step_avg:211.47ms
step:3743/4740 train_time:791517ms step_avg:211.47ms
step:3744/4740 train_time:791838ms step_avg:211.50ms
step:3745/4740 train_time:792015ms step_avg:211.49ms
step:3746/4740 train_time:792338ms step_avg:211.52ms
step:3747/4740 train_time:792518ms step_avg:211.51ms
step:3748/4740 train_time:792840ms step_avg:211.54ms
step:3749/4740 train_time:793015ms step_avg:211.53ms
step:3750/4740 train_time:793337ms step_avg:211.56ms
step:3750/4740 val_loss:3.0082 train_time:793460ms step_avg:211.59ms
step:3751/4740 train_time:793518ms step_avg:211.55ms
step:3752/4740 train_time:793848ms step_avg:211.58ms
step:3753/4740 train_time:794027ms step_avg:211.57ms
step:3754/4740 train_time:794343ms step_avg:211.60ms
step:3755/4740 train_time:794526ms step_avg:211.59ms
step:3756/4740 train_time:794850ms step_avg:211.62ms
step:3757/4740 train_time:795032ms step_avg:211.61ms
step:3758/4740 train_time:795352ms step_avg:211.64ms
step:3759/4740 train_time:795531ms step_avg:211.63ms
step:3760/4740 train_time:795859ms step_avg:211.66ms
step:3761/4740 train_time:796037ms step_avg:211.66ms
step:3762/4740 train_time:796355ms step_avg:211.68ms
step:3763/4740 train_time:796531ms step_avg:211.67ms
step:3764/4740 train_time:796852ms step_avg:211.70ms
step:3765/4740 train_time:797028ms step_avg:211.69ms
step:3766/4740 train_time:797348ms step_avg:211.72ms
step:3767/4740 train_time:797527ms step_avg:211.71ms
step:3768/4740 train_time:797847ms step_avg:211.74ms
step:3769/4740 train_time:798026ms step_avg:211.73ms
step:3770/4740 train_time:798347ms step_avg:211.76ms
step:3771/4740 train_time:798526ms step_avg:211.75ms
step:3772/4740 train_time:798846ms step_avg:211.78ms
step:3773/4740 train_time:799024ms step_avg:211.77ms
step:3774/4740 train_time:799347ms step_avg:211.80ms
step:3775/4740 train_time:799525ms step_avg:211.79ms
step:3776/4740 train_time:799846ms step_avg:211.82ms
step:3777/4740 train_time:800023ms step_avg:211.81ms
step:3778/4740 train_time:800344ms step_avg:211.84ms
step:3779/4740 train_time:800521ms step_avg:211.83ms
step:3780/4740 train_time:800843ms step_avg:211.86ms
step:3781/4740 train_time:801021ms step_avg:211.85ms
step:3782/4740 train_time:801343ms step_avg:211.88ms
step:3783/4740 train_time:801519ms step_avg:211.87ms
step:3784/4740 train_time:801837ms step_avg:211.90ms
step:3785/4740 train_time:802015ms step_avg:211.89ms
step:3786/4740 train_time:802335ms step_avg:211.92ms
step:3787/4740 train_time:802516ms step_avg:211.91ms
step:3788/4740 train_time:802835ms step_avg:211.94ms
step:3789/4740 train_time:803015ms step_avg:211.93ms
step:3790/4740 train_time:803334ms step_avg:211.96ms
step:3791/4740 train_time:803512ms step_avg:211.95ms
step:3792/4740 train_time:803828ms step_avg:211.98ms
step:3793/4740 train_time:804010ms step_avg:211.97ms
step:3794/4740 train_time:804330ms step_avg:212.00ms
step:3795/4740 train_time:804509ms step_avg:211.99ms
step:3796/4740 train_time:804830ms step_avg:212.02ms
step:3797/4740 train_time:805009ms step_avg:212.01ms
step:3798/4740 train_time:805332ms step_avg:212.04ms
step:3799/4740 train_time:805512ms step_avg:212.03ms
step:3800/4740 train_time:805832ms step_avg:212.06ms
step:3801/4740 train_time:806009ms step_avg:212.05ms
step:3802/4740 train_time:806329ms step_avg:212.08ms
step:3803/4740 train_time:806505ms step_avg:212.07ms
step:3804/4740 train_time:806830ms step_avg:212.10ms
step:3805/4740 train_time:807007ms step_avg:212.09ms
step:3806/4740 train_time:807330ms step_avg:212.12ms
step:3807/4740 train_time:807506ms step_avg:212.11ms
step:3808/4740 train_time:807828ms step_avg:212.14ms
step:3809/4740 train_time:808006ms step_avg:212.13ms
step:3810/4740 train_time:808329ms step_avg:212.16ms
step:3811/4740 train_time:808509ms step_avg:212.15ms
step:3812/4740 train_time:808830ms step_avg:212.18ms
step:3813/4740 train_time:809007ms step_avg:212.17ms
step:3814/4740 train_time:809330ms step_avg:212.20ms
step:3815/4740 train_time:809505ms step_avg:212.19ms
step:3816/4740 train_time:809827ms step_avg:212.22ms
step:3817/4740 train_time:810006ms step_avg:212.21ms
step:3818/4740 train_time:810327ms step_avg:212.24ms
step:3819/4740 train_time:810505ms step_avg:212.23ms
step:3820/4740 train_time:810830ms step_avg:212.26ms
step:3821/4740 train_time:811005ms step_avg:212.25ms
step:3822/4740 train_time:811327ms step_avg:212.28ms
step:3823/4740 train_time:811506ms step_avg:212.27ms
step:3824/4740 train_time:811831ms step_avg:212.30ms
step:3825/4740 train_time:812008ms step_avg:212.29ms
step:3826/4740 train_time:812331ms step_avg:212.32ms
step:3827/4740 train_time:812510ms step_avg:212.31ms
step:3828/4740 train_time:812833ms step_avg:212.34ms
step:3829/4740 train_time:813008ms step_avg:212.33ms
step:3830/4740 train_time:813328ms step_avg:212.36ms
step:3831/4740 train_time:813508ms step_avg:212.35ms
step:3832/4740 train_time:813828ms step_avg:212.38ms
step:3833/4740 train_time:814007ms step_avg:212.37ms
step:3834/4740 train_time:814327ms step_avg:212.40ms
step:3835/4740 train_time:814505ms step_avg:212.39ms
step:3836/4740 train_time:814824ms step_avg:212.42ms
step:3837/4740 train_time:815002ms step_avg:212.41ms
step:3838/4740 train_time:815327ms step_avg:212.44ms
step:3839/4740 train_time:815503ms step_avg:212.43ms
step:3840/4740 train_time:815821ms step_avg:212.45ms
step:3841/4740 train_time:815999ms step_avg:212.44ms
step:3842/4740 train_time:816320ms step_avg:212.47ms
step:3843/4740 train_time:816497ms step_avg:212.46ms
step:3844/4740 train_time:816821ms step_avg:212.49ms
step:3845/4740 train_time:816997ms step_avg:212.48ms
step:3846/4740 train_time:817321ms step_avg:212.51ms
step:3847/4740 train_time:817499ms step_avg:212.50ms
step:3848/4740 train_time:817824ms step_avg:212.53ms
step:3849/4740 train_time:818000ms step_avg:212.52ms
step:3850/4740 train_time:818317ms step_avg:212.55ms
step:3851/4740 train_time:818495ms step_avg:212.54ms
step:3852/4740 train_time:818817ms step_avg:212.57ms
step:3853/4740 train_time:818994ms step_avg:212.56ms
step:3854/4740 train_time:819317ms step_avg:212.59ms
step:3855/4740 train_time:819497ms step_avg:212.58ms
step:3856/4740 train_time:819817ms step_avg:212.61ms
step:3857/4740 train_time:819994ms step_avg:212.60ms
step:3858/4740 train_time:820314ms step_avg:212.63ms
step:3859/4740 train_time:820496ms step_avg:212.62ms
step:3860/4740 train_time:820818ms step_avg:212.65ms
step:3861/4740 train_time:820998ms step_avg:212.64ms
step:3862/4740 train_time:821318ms step_avg:212.67ms
step:3863/4740 train_time:821499ms step_avg:212.66ms
step:3864/4740 train_time:821817ms step_avg:212.69ms
step:3865/4740 train_time:821996ms step_avg:212.68ms
step:3866/4740 train_time:822319ms step_avg:212.71ms
step:3867/4740 train_time:822496ms step_avg:212.70ms
step:3868/4740 train_time:822817ms step_avg:212.72ms
step:3869/4740 train_time:822994ms step_avg:212.71ms
step:3870/4740 train_time:823315ms step_avg:212.74ms
step:3871/4740 train_time:823491ms step_avg:212.73ms
step:3872/4740 train_time:823815ms step_avg:212.76ms
step:3873/4740 train_time:823995ms step_avg:212.75ms
step:3874/4740 train_time:824316ms step_avg:212.78ms
step:3875/4740 train_time:824499ms step_avg:212.77ms
step:3876/4740 train_time:824823ms step_avg:212.80ms
step:3877/4740 train_time:825000ms step_avg:212.79ms
step:3878/4740 train_time:825320ms step_avg:212.82ms
step:3879/4740 train_time:825499ms step_avg:212.81ms
step:3880/4740 train_time:825814ms step_avg:212.84ms
step:3881/4740 train_time:825996ms step_avg:212.83ms
step:3882/4740 train_time:826316ms step_avg:212.86ms
step:3883/4740 train_time:826495ms step_avg:212.85ms
step:3884/4740 train_time:826815ms step_avg:212.88ms
step:3885/4740 train_time:826997ms step_avg:212.87ms
step:3886/4740 train_time:827317ms step_avg:212.90ms
step:3887/4740 train_time:827493ms step_avg:212.89ms
step:3888/4740 train_time:827814ms step_avg:212.92ms
step:3889/4740 train_time:827996ms step_avg:212.91ms
step:3890/4740 train_time:828313ms step_avg:212.93ms
step:3891/4740 train_time:828492ms step_avg:212.93ms
step:3892/4740 train_time:828814ms step_avg:212.95ms
step:3893/4740 train_time:828993ms step_avg:212.94ms
step:3894/4740 train_time:829317ms step_avg:212.97ms
step:3895/4740 train_time:829495ms step_avg:212.96ms
step:3896/4740 train_time:829821ms step_avg:212.99ms
step:3897/4740 train_time:829997ms step_avg:212.98ms
step:3898/4740 train_time:830319ms step_avg:213.01ms
step:3899/4740 train_time:830498ms step_avg:213.00ms
step:3900/4740 train_time:830822ms step_avg:213.03ms
step:3901/4740 train_time:830998ms step_avg:213.02ms
step:3902/4740 train_time:831321ms step_avg:213.05ms
step:3903/4740 train_time:831499ms step_avg:213.04ms
step:3904/4740 train_time:831821ms step_avg:213.07ms
step:3905/4740 train_time:832002ms step_avg:213.06ms
step:3906/4740 train_time:832324ms step_avg:213.09ms
step:3907/4740 train_time:832507ms step_avg:213.08ms
step:3908/4740 train_time:832832ms step_avg:213.11ms
step:3909/4740 train_time:833009ms step_avg:213.10ms
step:3910/4740 train_time:833330ms step_avg:213.13ms
step:3911/4740 train_time:833507ms step_avg:213.12ms
step:3912/4740 train_time:833828ms step_avg:213.15ms
step:3913/4740 train_time:834007ms step_avg:213.14ms
step:3914/4740 train_time:834328ms step_avg:213.17ms
step:3915/4740 train_time:834505ms step_avg:213.16ms
step:3916/4740 train_time:834830ms step_avg:213.18ms
step:3917/4740 train_time:835009ms step_avg:213.18ms
step:3918/4740 train_time:835330ms step_avg:213.20ms
step:3919/4740 train_time:835509ms step_avg:213.19ms
step:3920/4740 train_time:835830ms step_avg:213.22ms
step:3921/4740 train_time:836009ms step_avg:213.21ms
step:3922/4740 train_time:836328ms step_avg:213.24ms
step:3923/4740 train_time:836507ms step_avg:213.23ms
step:3924/4740 train_time:836834ms step_avg:213.26ms
step:3925/4740 train_time:837010ms step_avg:213.25ms
step:3926/4740 train_time:837330ms step_avg:213.28ms
step:3927/4740 train_time:837508ms step_avg:213.27ms
step:3928/4740 train_time:837829ms step_avg:213.30ms
step:3929/4740 train_time:838006ms step_avg:213.29ms
step:3930/4740 train_time:838329ms step_avg:213.32ms
step:3931/4740 train_time:838507ms step_avg:213.31ms
step:3932/4740 train_time:838830ms step_avg:213.33ms
step:3933/4740 train_time:839005ms step_avg:213.32ms
step:3934/4740 train_time:839330ms step_avg:213.35ms
step:3935/4740 train_time:839510ms step_avg:213.34ms
step:3936/4740 train_time:839834ms step_avg:213.37ms
step:3937/4740 train_time:840009ms step_avg:213.36ms
step:3938/4740 train_time:840333ms step_avg:213.39ms
step:3939/4740 train_time:840511ms step_avg:213.38ms
step:3940/4740 train_time:840830ms step_avg:213.41ms
step:3941/4740 train_time:841011ms step_avg:213.40ms
step:3942/4740 train_time:841334ms step_avg:213.43ms
step:3943/4740 train_time:841515ms step_avg:213.42ms
step:3944/4740 train_time:841835ms step_avg:213.45ms
step:3945/4740 train_time:842017ms step_avg:213.44ms
step:3946/4740 train_time:842339ms step_avg:213.47ms
step:3947/4740 train_time:842521ms step_avg:213.46ms
step:3948/4740 train_time:842843ms step_avg:213.49ms
step:3949/4740 train_time:843021ms step_avg:213.48ms
step:3950/4740 train_time:843345ms step_avg:213.50ms
step:3951/4740 train_time:843521ms step_avg:213.50ms
step:3952/4740 train_time:843840ms step_avg:213.52ms
step:3953/4740 train_time:844018ms step_avg:213.51ms
step:3954/4740 train_time:844341ms step_avg:213.54ms
step:3955/4740 train_time:844518ms step_avg:213.53ms
step:3956/4740 train_time:844838ms step_avg:213.56ms
step:3957/4740 train_time:845017ms step_avg:213.55ms
step:3958/4740 train_time:845342ms step_avg:213.58ms
step:3959/4740 train_time:845526ms step_avg:213.57ms
step:3960/4740 train_time:845847ms step_avg:213.60ms
step:3961/4740 train_time:846028ms step_avg:213.59ms
step:3962/4740 train_time:846351ms step_avg:213.62ms
step:3963/4740 train_time:846531ms step_avg:213.61ms
step:3964/4740 train_time:846852ms step_avg:213.64ms
step:3965/4740 train_time:847030ms step_avg:213.63ms
step:3966/4740 train_time:847351ms step_avg:213.65ms
step:3967/4740 train_time:847529ms step_avg:213.64ms
step:3968/4740 train_time:847851ms step_avg:213.67ms
step:3969/4740 train_time:848028ms step_avg:213.66ms
step:3970/4740 train_time:848350ms step_avg:213.69ms
step:3971/4740 train_time:848528ms step_avg:213.68ms
step:3972/4740 train_time:848850ms step_avg:213.71ms
step:3973/4740 train_time:849027ms step_avg:213.70ms
step:3974/4740 train_time:849346ms step_avg:213.73ms
step:3975/4740 train_time:849524ms step_avg:213.72ms
step:3976/4740 train_time:849846ms step_avg:213.74ms
step:3977/4740 train_time:850025ms step_avg:213.74ms
step:3978/4740 train_time:850346ms step_avg:213.76ms
step:3979/4740 train_time:850524ms step_avg:213.75ms
step:3980/4740 train_time:850846ms step_avg:213.78ms
step:3981/4740 train_time:851025ms step_avg:213.77ms
step:3982/4740 train_time:851346ms step_avg:213.80ms
step:3983/4740 train_time:851526ms step_avg:213.79ms
step:3984/4740 train_time:851846ms step_avg:213.82ms
step:3985/4740 train_time:852024ms step_avg:213.81ms
step:3986/4740 train_time:852344ms step_avg:213.83ms
step:3987/4740 train_time:852524ms step_avg:213.83ms
step:3988/4740 train_time:852845ms step_avg:213.85ms
step:3989/4740 train_time:853026ms step_avg:213.84ms
step:3990/4740 train_time:853346ms step_avg:213.87ms
step:3991/4740 train_time:853526ms step_avg:213.86ms
step:3992/4740 train_time:853848ms step_avg:213.89ms
step:3993/4740 train_time:854026ms step_avg:213.88ms
step:3994/4740 train_time:854346ms step_avg:213.91ms
step:3995/4740 train_time:854527ms step_avg:213.90ms
step:3996/4740 train_time:854850ms step_avg:213.93ms
step:3997/4740 train_time:855028ms step_avg:213.92ms
step:3998/4740 train_time:855350ms step_avg:213.94ms
step:3999/4740 train_time:855526ms step_avg:213.94ms
step:4000/4740 train_time:855850ms step_avg:213.96ms
step:4000/4740 val_loss:2.9816 train_time:855971ms step_avg:213.99ms
step:4001/4740 train_time:856028ms step_avg:213.95ms
step:4002/4740 train_time:856358ms step_avg:213.98ms
step:4003/4740 train_time:856533ms step_avg:213.97ms
step:4004/4740 train_time:856848ms step_avg:214.00ms
step:4005/4740 train_time:857025ms step_avg:213.99ms
step:4006/4740 train_time:857354ms step_avg:214.02ms
step:4007/4740 train_time:857532ms step_avg:214.01ms
step:4008/4740 train_time:857852ms step_avg:214.04ms
step:4009/4740 train_time:858026ms step_avg:214.03ms
step:4010/4740 train_time:858354ms step_avg:214.05ms
step:4011/4740 train_time:858530ms step_avg:214.04ms
step:4012/4740 train_time:858851ms step_avg:214.07ms
step:4013/4740 train_time:859027ms step_avg:214.06ms
step:4014/4740 train_time:859351ms step_avg:214.09ms
step:4015/4740 train_time:859537ms step_avg:214.08ms
step:4016/4740 train_time:859857ms step_avg:214.11ms
step:4017/4740 train_time:860032ms step_avg:214.10ms
step:4018/4740 train_time:860358ms step_avg:214.13ms
step:4019/4740 train_time:860536ms step_avg:214.12ms
step:4020/4740 train_time:860859ms step_avg:214.14ms
step:4021/4740 train_time:861035ms step_avg:214.13ms
step:4022/4740 train_time:861359ms step_avg:214.16ms
step:4023/4740 train_time:861535ms step_avg:214.15ms
step:4024/4740 train_time:861854ms step_avg:214.18ms
step:4025/4740 train_time:862034ms step_avg:214.17ms
step:4026/4740 train_time:862353ms step_avg:214.20ms
step:4027/4740 train_time:862533ms step_avg:214.19ms
step:4028/4740 train_time:862855ms step_avg:214.21ms
step:4029/4740 train_time:863033ms step_avg:214.21ms
step:4030/4740 train_time:863355ms step_avg:214.23ms
step:4031/4740 train_time:863533ms step_avg:214.22ms
step:4032/4740 train_time:863854ms step_avg:214.25ms
step:4033/4740 train_time:864035ms step_avg:214.24ms
step:4034/4740 train_time:864356ms step_avg:214.27ms
step:4035/4740 train_time:864538ms step_avg:214.26ms
step:4036/4740 train_time:864856ms step_avg:214.29ms
step:4037/4740 train_time:865037ms step_avg:214.28ms
step:4038/4740 train_time:865361ms step_avg:214.30ms
step:4039/4740 train_time:865541ms step_avg:214.30ms
step:4040/4740 train_time:865859ms step_avg:214.32ms
step:4041/4740 train_time:866039ms step_avg:214.31ms
step:4042/4740 train_time:866358ms step_avg:214.34ms
step:4043/4740 train_time:866537ms step_avg:214.33ms
step:4044/4740 train_time:866857ms step_avg:214.36ms
step:4045/4740 train_time:867034ms step_avg:214.35ms
step:4046/4740 train_time:867352ms step_avg:214.37ms
step:4047/4740 train_time:867533ms step_avg:214.36ms
step:4048/4740 train_time:867852ms step_avg:214.39ms
step:4049/4740 train_time:868030ms step_avg:214.38ms
step:4050/4740 train_time:868353ms step_avg:214.41ms
step:4051/4740 train_time:868531ms step_avg:214.40ms
step:4052/4740 train_time:868850ms step_avg:214.43ms
step:4053/4740 train_time:869030ms step_avg:214.42ms
step:4054/4740 train_time:869352ms step_avg:214.44ms
step:4055/4740 train_time:869528ms step_avg:214.43ms
step:4056/4740 train_time:869848ms step_avg:214.46ms
step:4057/4740 train_time:870028ms step_avg:214.45ms
step:4058/4740 train_time:870350ms step_avg:214.48ms
step:4059/4740 train_time:870527ms step_avg:214.47ms
step:4060/4740 train_time:870848ms step_avg:214.49ms
step:4061/4740 train_time:871027ms step_avg:214.49ms
step:4062/4740 train_time:871344ms step_avg:214.51ms
step:4063/4740 train_time:871526ms step_avg:214.50ms
step:4064/4740 train_time:871849ms step_avg:214.53ms
step:4065/4740 train_time:872026ms step_avg:214.52ms
step:4066/4740 train_time:872348ms step_avg:214.55ms
step:4067/4740 train_time:872526ms step_avg:214.54ms
step:4068/4740 train_time:872844ms step_avg:214.56ms
step:4069/4740 train_time:873023ms step_avg:214.55ms
step:4070/4740 train_time:873342ms step_avg:214.58ms
step:4071/4740 train_time:873524ms step_avg:214.57ms
step:4072/4740 train_time:873847ms step_avg:214.60ms
step:4073/4740 train_time:874024ms step_avg:214.59ms
step:4074/4740 train_time:874342ms step_avg:214.62ms
step:4075/4740 train_time:874526ms step_avg:214.61ms
step:4076/4740 train_time:874848ms step_avg:214.63ms
step:4077/4740 train_time:875029ms step_avg:214.63ms
step:4078/4740 train_time:875350ms step_avg:214.65ms
step:4079/4740 train_time:875529ms step_avg:214.64ms
step:4080/4740 train_time:875850ms step_avg:214.67ms
step:4081/4740 train_time:876029ms step_avg:214.66ms
step:4082/4740 train_time:876350ms step_avg:214.69ms
step:4083/4740 train_time:876527ms step_avg:214.68ms
step:4084/4740 train_time:876848ms step_avg:214.70ms
step:4085/4740 train_time:877025ms step_avg:214.69ms
step:4086/4740 train_time:877347ms step_avg:214.72ms
step:4087/4740 train_time:877526ms step_avg:214.71ms
step:4088/4740 train_time:877848ms step_avg:214.74ms
step:4089/4740 train_time:878026ms step_avg:214.73ms
step:4090/4740 train_time:878347ms step_avg:214.75ms
step:4091/4740 train_time:878524ms step_avg:214.75ms
step:4092/4740 train_time:878842ms step_avg:214.77ms
step:4093/4740 train_time:879024ms step_avg:214.76ms
step:4094/4740 train_time:879344ms step_avg:214.79ms
step:4095/4740 train_time:879523ms step_avg:214.78ms
step:4096/4740 train_time:879847ms step_avg:214.81ms
step:4097/4740 train_time:880024ms step_avg:214.80ms
step:4098/4740 train_time:880344ms step_avg:214.82ms
step:4099/4740 train_time:880522ms step_avg:214.81ms
step:4100/4740 train_time:880842ms step_avg:214.84ms
step:4101/4740 train_time:881021ms step_avg:214.83ms
step:4102/4740 train_time:881341ms step_avg:214.86ms
step:4103/4740 train_time:881520ms step_avg:214.85ms
step:4104/4740 train_time:881840ms step_avg:214.87ms
step:4105/4740 train_time:882019ms step_avg:214.86ms
step:4106/4740 train_time:882335ms step_avg:214.89ms
step:4107/4740 train_time:882515ms step_avg:214.88ms
step:4108/4740 train_time:882837ms step_avg:214.91ms
step:4109/4740 train_time:883018ms step_avg:214.90ms
step:4110/4740 train_time:883339ms step_avg:214.92ms
step:4111/4740 train_time:883517ms step_avg:214.92ms
step:4112/4740 train_time:883839ms step_avg:214.94ms
step:4113/4740 train_time:884017ms step_avg:214.93ms
step:4114/4740 train_time:884337ms step_avg:214.96ms
step:4115/4740 train_time:884516ms step_avg:214.95ms
step:4116/4740 train_time:884840ms step_avg:214.98ms
step:4117/4740 train_time:885017ms step_avg:214.97ms
step:4118/4740 train_time:885341ms step_avg:214.99ms
step:4119/4740 train_time:885521ms step_avg:214.98ms
step:4120/4740 train_time:885841ms step_avg:215.01ms
step:4121/4740 train_time:886018ms step_avg:215.00ms
step:4122/4740 train_time:886340ms step_avg:215.03ms
step:4123/4740 train_time:886519ms step_avg:215.02ms
step:4124/4740 train_time:886841ms step_avg:215.04ms
step:4125/4740 train_time:887016ms step_avg:215.03ms
step:4126/4740 train_time:887339ms step_avg:215.06ms
step:4127/4740 train_time:887518ms step_avg:215.05ms
step:4128/4740 train_time:887838ms step_avg:215.08ms
step:4129/4740 train_time:888021ms step_avg:215.07ms
step:4130/4740 train_time:888342ms step_avg:215.10ms
step:4131/4740 train_time:888523ms step_avg:215.09ms
step:4132/4740 train_time:888840ms step_avg:215.11ms
step:4133/4740 train_time:889024ms step_avg:215.10ms
step:4134/4740 train_time:889343ms step_avg:215.13ms
step:4135/4740 train_time:889528ms step_avg:215.12ms
step:4136/4740 train_time:889849ms step_avg:215.15ms
step:4137/4740 train_time:890029ms step_avg:215.14ms
step:4138/4740 train_time:890350ms step_avg:215.16ms
step:4139/4740 train_time:890527ms step_avg:215.16ms
step:4140/4740 train_time:890851ms step_avg:215.18ms
step:4141/4740 train_time:891027ms step_avg:215.17ms
step:4142/4740 train_time:891345ms step_avg:215.20ms
step:4143/4740 train_time:891524ms step_avg:215.19ms
step:4144/4740 train_time:891849ms step_avg:215.21ms
step:4145/4740 train_time:892028ms step_avg:215.21ms
step:4146/4740 train_time:892349ms step_avg:215.23ms
step:4147/4740 train_time:892530ms step_avg:215.22ms
step:4148/4740 train_time:892851ms step_avg:215.25ms
step:4149/4740 train_time:893026ms step_avg:215.24ms
step:4150/4740 train_time:893347ms step_avg:215.26ms
step:4151/4740 train_time:893528ms step_avg:215.26ms
step:4152/4740 train_time:893848ms step_avg:215.28ms
step:4153/4740 train_time:894026ms step_avg:215.27ms
step:4154/4740 train_time:894349ms step_avg:215.30ms
step:4155/4740 train_time:894525ms step_avg:215.29ms
step:4156/4740 train_time:894848ms step_avg:215.31ms
step:4157/4740 train_time:895028ms step_avg:215.31ms
step:4158/4740 train_time:895347ms step_avg:215.33ms
step:4159/4740 train_time:895524ms step_avg:215.32ms
step:4160/4740 train_time:895848ms step_avg:215.35ms
step:4161/4740 train_time:896027ms step_avg:215.34ms
step:4162/4740 train_time:896348ms step_avg:215.36ms
step:4163/4740 train_time:896526ms step_avg:215.36ms
step:4164/4740 train_time:896848ms step_avg:215.38ms
step:4165/4740 train_time:897029ms step_avg:215.37ms
step:4166/4740 train_time:897350ms step_avg:215.40ms
step:4167/4740 train_time:897528ms step_avg:215.39ms
step:4168/4740 train_time:897850ms step_avg:215.42ms
step:4169/4740 train_time:898028ms step_avg:215.41ms
step:4170/4740 train_time:898348ms step_avg:215.43ms
step:4171/4740 train_time:898530ms step_avg:215.42ms
step:4172/4740 train_time:898853ms step_avg:215.45ms
step:4173/4740 train_time:899033ms step_avg:215.44ms
step:4174/4740 train_time:899353ms step_avg:215.47ms
step:4175/4740 train_time:899532ms step_avg:215.46ms
step:4176/4740 train_time:899855ms step_avg:215.48ms
step:4177/4740 train_time:900033ms step_avg:215.47ms
step:4178/4740 train_time:900352ms step_avg:215.50ms
step:4179/4740 train_time:900529ms step_avg:215.49ms
step:4180/4740 train_time:900852ms step_avg:215.51ms
step:4181/4740 train_time:901030ms step_avg:215.51ms
step:4182/4740 train_time:901352ms step_avg:215.53ms
step:4183/4740 train_time:901528ms step_avg:215.52ms
step:4184/4740 train_time:901851ms step_avg:215.55ms
step:4185/4740 train_time:902030ms step_avg:215.54ms
step:4186/4740 train_time:902354ms step_avg:215.56ms
step:4187/4740 train_time:902530ms step_avg:215.56ms
step:4188/4740 train_time:902850ms step_avg:215.58ms
step:4189/4740 train_time:903030ms step_avg:215.57ms
step:4190/4740 train_time:903351ms step_avg:215.60ms
step:4191/4740 train_time:903529ms step_avg:215.59ms
step:4192/4740 train_time:903851ms step_avg:215.61ms
step:4193/4740 train_time:904028ms step_avg:215.60ms
step:4194/4740 train_time:904348ms step_avg:215.63ms
step:4195/4740 train_time:904526ms step_avg:215.62ms
step:4196/4740 train_time:904847ms step_avg:215.65ms
step:4197/4740 train_time:905028ms step_avg:215.64ms
step:4198/4740 train_time:905349ms step_avg:215.66ms
step:4199/4740 train_time:905529ms step_avg:215.65ms
step:4200/4740 train_time:905851ms step_avg:215.68ms
step:4201/4740 train_time:906029ms step_avg:215.67ms
step:4202/4740 train_time:906352ms step_avg:215.70ms
step:4203/4740 train_time:906532ms step_avg:215.69ms
step:4204/4740 train_time:906855ms step_avg:215.71ms
step:4205/4740 train_time:907032ms step_avg:215.70ms
step:4206/4740 train_time:907352ms step_avg:215.73ms
step:4207/4740 train_time:907530ms step_avg:215.72ms
step:4208/4740 train_time:907850ms step_avg:215.74ms
step:4209/4740 train_time:908034ms step_avg:215.74ms
step:4210/4740 train_time:908356ms step_avg:215.76ms
step:4211/4740 train_time:908532ms step_avg:215.75ms
step:4212/4740 train_time:908852ms step_avg:215.78ms
step:4213/4740 train_time:909032ms step_avg:215.77ms
step:4214/4740 train_time:909353ms step_avg:215.79ms
step:4215/4740 train_time:909532ms step_avg:215.78ms
step:4216/4740 train_time:909853ms step_avg:215.81ms
step:4217/4740 train_time:910031ms step_avg:215.80ms
step:4218/4740 train_time:910353ms step_avg:215.83ms
step:4219/4740 train_time:910531ms step_avg:215.82ms
step:4220/4740 train_time:910853ms step_avg:215.84ms
step:4221/4740 train_time:911031ms step_avg:215.83ms
step:4222/4740 train_time:911351ms step_avg:215.86ms
step:4223/4740 train_time:911531ms step_avg:215.85ms
step:4224/4740 train_time:911854ms step_avg:215.87ms
step:4225/4740 train_time:912036ms step_avg:215.87ms
step:4226/4740 train_time:912357ms step_avg:215.89ms
step:4227/4740 train_time:912537ms step_avg:215.88ms
step:4228/4740 train_time:912858ms step_avg:215.91ms
step:4229/4740 train_time:913038ms step_avg:215.90ms
step:4230/4740 train_time:913353ms step_avg:215.92ms
step:4231/4740 train_time:913535ms step_avg:215.91ms
step:4232/4740 train_time:913856ms step_avg:215.94ms
step:4233/4740 train_time:914036ms step_avg:215.93ms
step:4234/4740 train_time:914355ms step_avg:215.96ms
step:4235/4740 train_time:914534ms step_avg:215.95ms
step:4236/4740 train_time:914857ms step_avg:215.97ms
step:4237/4740 train_time:915035ms step_avg:215.96ms
step:4238/4740 train_time:915354ms step_avg:215.99ms
step:4239/4740 train_time:915533ms step_avg:215.98ms
step:4240/4740 train_time:915858ms step_avg:216.00ms
step:4241/4740 train_time:916038ms step_avg:216.00ms
step:4242/4740 train_time:916359ms step_avg:216.02ms
step:4243/4740 train_time:916534ms step_avg:216.01ms
step:4244/4740 train_time:916852ms step_avg:216.03ms
step:4245/4740 train_time:917034ms step_avg:216.03ms
step:4246/4740 train_time:917355ms step_avg:216.05ms
step:4247/4740 train_time:917534ms step_avg:216.04ms
step:4248/4740 train_time:917855ms step_avg:216.07ms
step:4249/4740 train_time:918034ms step_avg:216.06ms
step:4250/4740 train_time:918353ms step_avg:216.08ms
step:4250/4740 val_loss:2.9560 train_time:918481ms step_avg:216.11ms
step:4251/4740 train_time:918538ms step_avg:216.08ms
step:4252/4740 train_time:918872ms step_avg:216.10ms
step:4253/4740 train_time:919051ms step_avg:216.09ms
step:4254/4740 train_time:919366ms step_avg:216.12ms
step:4255/4740 train_time:919544ms step_avg:216.11ms
step:4256/4740 train_time:919868ms step_avg:216.13ms
step:4257/4740 train_time:920050ms step_avg:216.13ms
step:4258/4740 train_time:920369ms step_avg:216.15ms
step:4259/4740 train_time:920545ms step_avg:216.14ms
step:4260/4740 train_time:920872ms step_avg:216.17ms
step:4261/4740 train_time:921052ms step_avg:216.16ms
step:4262/4740 train_time:921368ms step_avg:216.18ms
step:4263/4740 train_time:921547ms step_avg:216.17ms
step:4264/4740 train_time:921867ms step_avg:216.20ms
step:4265/4740 train_time:922048ms step_avg:216.19ms
step:4266/4740 train_time:922367ms step_avg:216.21ms
step:4267/4740 train_time:922542ms step_avg:216.20ms
step:4268/4740 train_time:922867ms step_avg:216.23ms
step:4269/4740 train_time:923050ms step_avg:216.22ms
step:4270/4740 train_time:923371ms step_avg:216.25ms
step:4271/4740 train_time:923547ms step_avg:216.24ms
step:4272/4740 train_time:923869ms step_avg:216.26ms
step:4273/4740 train_time:924050ms step_avg:216.25ms
step:4274/4740 train_time:924373ms step_avg:216.28ms
step:4275/4740 train_time:924551ms step_avg:216.27ms
step:4276/4740 train_time:924875ms step_avg:216.29ms
step:4277/4740 train_time:925051ms step_avg:216.29ms
step:4278/4740 train_time:925370ms step_avg:216.31ms
step:4279/4740 train_time:925549ms step_avg:216.30ms
step:4280/4740 train_time:925872ms step_avg:216.33ms
step:4281/4740 train_time:926051ms step_avg:216.32ms
step:4282/4740 train_time:926370ms step_avg:216.34ms
step:4283/4740 train_time:926548ms step_avg:216.33ms
step:4284/4740 train_time:926869ms step_avg:216.36ms
step:4285/4740 train_time:927050ms step_avg:216.35ms
step:4286/4740 train_time:927367ms step_avg:216.37ms
step:4287/4740 train_time:927549ms step_avg:216.36ms
step:4288/4740 train_time:927866ms step_avg:216.39ms
step:4289/4740 train_time:928046ms step_avg:216.38ms
step:4290/4740 train_time:928365ms step_avg:216.40ms
step:4291/4740 train_time:928545ms step_avg:216.39ms
step:4292/4740 train_time:928867ms step_avg:216.42ms
step:4293/4740 train_time:929048ms step_avg:216.41ms
step:4294/4740 train_time:929371ms step_avg:216.43ms
step:4295/4740 train_time:929551ms step_avg:216.43ms
step:4296/4740 train_time:929872ms step_avg:216.45ms
step:4297/4740 train_time:930050ms step_avg:216.44ms
step:4298/4740 train_time:930372ms step_avg:216.47ms
step:4299/4740 train_time:930554ms step_avg:216.46ms
step:4300/4740 train_time:930877ms step_avg:216.48ms
step:4301/4740 train_time:931057ms step_avg:216.47ms
step:4302/4740 train_time:931379ms step_avg:216.50ms
step:4303/4740 train_time:931560ms step_avg:216.49ms
step:4304/4740 train_time:931883ms step_avg:216.52ms
step:4305/4740 train_time:932063ms step_avg:216.51ms
step:4306/4740 train_time:932386ms step_avg:216.53ms
step:4307/4740 train_time:932563ms step_avg:216.52ms
step:4308/4740 train_time:932885ms step_avg:216.55ms
step:4309/4740 train_time:933063ms step_avg:216.54ms
step:4310/4740 train_time:933383ms step_avg:216.56ms
step:4311/4740 train_time:933560ms step_avg:216.55ms
step:4312/4740 train_time:933885ms step_avg:216.58ms
step:4313/4740 train_time:934061ms step_avg:216.57ms
step:4314/4740 train_time:934382ms step_avg:216.59ms
step:4315/4740 train_time:934559ms step_avg:216.58ms
step:4316/4740 train_time:934885ms step_avg:216.61ms
step:4317/4740 train_time:935062ms step_avg:216.60ms
step:4318/4740 train_time:935386ms step_avg:216.62ms
step:4319/4740 train_time:935561ms step_avg:216.62ms
step:4320/4740 train_time:935884ms step_avg:216.64ms
step:4321/4740 train_time:936064ms step_avg:216.63ms
step:4322/4740 train_time:936384ms step_avg:216.66ms
step:4323/4740 train_time:936561ms step_avg:216.65ms
step:4324/4740 train_time:936886ms step_avg:216.67ms
step:4325/4740 train_time:937064ms step_avg:216.66ms
step:4326/4740 train_time:937384ms step_avg:216.69ms
step:4327/4740 train_time:937564ms step_avg:216.68ms
step:4328/4740 train_time:937887ms step_avg:216.70ms
step:4329/4740 train_time:938062ms step_avg:216.69ms
step:4330/4740 train_time:938384ms step_avg:216.72ms
step:4331/4740 train_time:938560ms step_avg:216.71ms
step:4332/4740 train_time:938885ms step_avg:216.73ms
step:4333/4740 train_time:939063ms step_avg:216.72ms
step:4334/4740 train_time:939383ms step_avg:216.75ms
step:4335/4740 train_time:939562ms step_avg:216.74ms
step:4336/4740 train_time:939884ms step_avg:216.76ms
step:4337/4740 train_time:940059ms step_avg:216.75ms
step:4338/4740 train_time:940382ms step_avg:216.78ms
step:4339/4740 train_time:940558ms step_avg:216.77ms
step:4340/4740 train_time:940880ms step_avg:216.79ms
step:4341/4740 train_time:941060ms step_avg:216.78ms
step:4342/4740 train_time:941383ms step_avg:216.81ms
step:4343/4740 train_time:941561ms step_avg:216.80ms
step:4344/4740 train_time:941885ms step_avg:216.82ms
step:4345/4740 train_time:942061ms step_avg:216.81ms
step:4346/4740 train_time:942382ms step_avg:216.84ms
step:4347/4740 train_time:942560ms step_avg:216.83ms
step:4348/4740 train_time:942883ms step_avg:216.85ms
step:4349/4740 train_time:943060ms step_avg:216.85ms
step:4350/4740 train_time:943384ms step_avg:216.87ms
step:4351/4740 train_time:943561ms step_avg:216.86ms
step:4352/4740 train_time:943883ms step_avg:216.88ms
step:4353/4740 train_time:944060ms step_avg:216.88ms
step:4354/4740 train_time:944384ms step_avg:216.90ms
step:4355/4740 train_time:944559ms step_avg:216.89ms
step:4356/4740 train_time:944882ms step_avg:216.92ms
step:4357/4740 train_time:945059ms step_avg:216.91ms
step:4358/4740 train_time:945384ms step_avg:216.93ms
step:4359/4740 train_time:945564ms step_avg:216.92ms
step:4360/4740 train_time:945884ms step_avg:216.95ms
step:4361/4740 train_time:946065ms step_avg:216.94ms
step:4362/4740 train_time:946388ms step_avg:216.96ms
step:4363/4740 train_time:946564ms step_avg:216.95ms
step:4364/4740 train_time:946889ms step_avg:216.98ms
step:4365/4740 train_time:947067ms step_avg:216.97ms
step:4366/4740 train_time:947388ms step_avg:216.99ms
step:4367/4740 train_time:947563ms step_avg:216.98ms
step:4368/4740 train_time:947889ms step_avg:217.01ms
step:4369/4740 train_time:948064ms step_avg:217.00ms
step:4370/4740 train_time:948385ms step_avg:217.02ms
step:4371/4740 train_time:948564ms step_avg:217.01ms
step:4372/4740 train_time:948886ms step_avg:217.04ms
step:4373/4740 train_time:949060ms step_avg:217.03ms
step:4374/4740 train_time:949383ms step_avg:217.05ms
step:4375/4740 train_time:949558ms step_avg:217.04ms
step:4376/4740 train_time:949884ms step_avg:217.07ms
step:4377/4740 train_time:950059ms step_avg:217.06ms
step:4378/4740 train_time:950380ms step_avg:217.08ms
step:4379/4740 train_time:950558ms step_avg:217.07ms
step:4380/4740 train_time:950882ms step_avg:217.10ms
step:4381/4740 train_time:951061ms step_avg:217.09ms
step:4382/4740 train_time:951382ms step_avg:217.11ms
step:4383/4740 train_time:951559ms step_avg:217.10ms
step:4384/4740 train_time:951882ms step_avg:217.13ms
step:4385/4740 train_time:952062ms step_avg:217.12ms
step:4386/4740 train_time:952384ms step_avg:217.14ms
step:4387/4740 train_time:952562ms step_avg:217.13ms
step:4388/4740 train_time:952881ms step_avg:217.16ms
step:4389/4740 train_time:953057ms step_avg:217.15ms
step:4390/4740 train_time:953380ms step_avg:217.17ms
step:4391/4740 train_time:953557ms step_avg:217.16ms
step:4392/4740 train_time:953881ms step_avg:217.19ms
step:4393/4740 train_time:954059ms step_avg:217.18ms
step:4394/4740 train_time:954379ms step_avg:217.20ms
step:4395/4740 train_time:954558ms step_avg:217.19ms
step:4396/4740 train_time:954880ms step_avg:217.22ms
step:4397/4740 train_time:955055ms step_avg:217.21ms
step:4398/4740 train_time:955378ms step_avg:217.23ms
step:4399/4740 train_time:955555ms step_avg:217.22ms
step:4400/4740 train_time:955877ms step_avg:217.24ms
step:4401/4740 train_time:956056ms step_avg:217.24ms
step:4402/4740 train_time:956381ms step_avg:217.26ms
step:4403/4740 train_time:956558ms step_avg:217.25ms
step:4404/4740 train_time:956881ms step_avg:217.28ms
step:4405/4740 train_time:957056ms step_avg:217.27ms
step:4406/4740 train_time:957380ms step_avg:217.29ms
step:4407/4740 train_time:957558ms step_avg:217.28ms
step:4408/4740 train_time:957882ms step_avg:217.31ms
step:4409/4740 train_time:958058ms step_avg:217.30ms
step:4410/4740 train_time:958380ms step_avg:217.32ms
step:4411/4740 train_time:958558ms step_avg:217.31ms
step:4412/4740 train_time:958878ms step_avg:217.33ms
step:4413/4740 train_time:959059ms step_avg:217.33ms
step:4414/4740 train_time:959378ms step_avg:217.35ms
step:4415/4740 train_time:959554ms step_avg:217.34ms
step:4416/4740 train_time:959877ms step_avg:217.36ms
step:4417/4740 train_time:960060ms step_avg:217.36ms
step:4418/4740 train_time:960380ms step_avg:217.38ms
step:4419/4740 train_time:960558ms step_avg:217.37ms
step:4420/4740 train_time:960884ms step_avg:217.39ms
step:4421/4740 train_time:961059ms step_avg:217.38ms
step:4422/4740 train_time:961379ms step_avg:217.41ms
step:4423/4740 train_time:961562ms step_avg:217.40ms
step:4424/4740 train_time:961885ms step_avg:217.42ms
step:4425/4740 train_time:962060ms step_avg:217.41ms
step:4426/4740 train_time:962380ms step_avg:217.44ms
step:4427/4740 train_time:962560ms step_avg:217.43ms
step:4428/4740 train_time:962884ms step_avg:217.45ms
step:4429/4740 train_time:963060ms step_avg:217.44ms
step:4430/4740 train_time:963380ms step_avg:217.47ms
step:4431/4740 train_time:963559ms step_avg:217.46ms
step:4432/4740 train_time:963885ms step_avg:217.48ms
step:4433/4740 train_time:964060ms step_avg:217.47ms
step:4434/4740 train_time:964380ms step_avg:217.50ms
step:4435/4740 train_time:964555ms step_avg:217.49ms
step:4436/4740 train_time:964879ms step_avg:217.51ms
step:4437/4740 train_time:965055ms step_avg:217.50ms
step:4438/4740 train_time:965381ms step_avg:217.53ms
step:4439/4740 train_time:965558ms step_avg:217.52ms
step:4440/4740 train_time:965882ms step_avg:217.54ms
step:4441/4740 train_time:966059ms step_avg:217.53ms
step:4442/4740 train_time:966380ms step_avg:217.56ms
step:4443/4740 train_time:966559ms step_avg:217.55ms
step:4444/4740 train_time:966879ms step_avg:217.57ms
step:4445/4740 train_time:967059ms step_avg:217.56ms
step:4446/4740 train_time:967380ms step_avg:217.58ms
step:4447/4740 train_time:967559ms step_avg:217.58ms
step:4448/4740 train_time:967880ms step_avg:217.60ms
step:4449/4740 train_time:968058ms step_avg:217.59ms
step:4450/4740 train_time:968380ms step_avg:217.61ms
step:4451/4740 train_time:968558ms step_avg:217.60ms
step:4452/4740 train_time:968881ms step_avg:217.63ms
step:4453/4740 train_time:969056ms step_avg:217.62ms
step:4454/4740 train_time:969381ms step_avg:217.64ms
step:4455/4740 train_time:969558ms step_avg:217.63ms
step:4456/4740 train_time:969880ms step_avg:217.66ms
step:4457/4740 train_time:970057ms step_avg:217.65ms
step:4458/4740 train_time:970382ms step_avg:217.67ms
step:4459/4740 train_time:970560ms step_avg:217.66ms
step:4460/4740 train_time:970881ms step_avg:217.69ms
step:4461/4740 train_time:971059ms step_avg:217.68ms
step:4462/4740 train_time:971380ms step_avg:217.70ms
step:4463/4740 train_time:971561ms step_avg:217.69ms
step:4464/4740 train_time:971884ms step_avg:217.72ms
step:4465/4740 train_time:972061ms step_avg:217.71ms
step:4466/4740 train_time:972385ms step_avg:217.73ms
step:4467/4740 train_time:972562ms step_avg:217.72ms
step:4468/4740 train_time:972884ms step_avg:217.74ms
step:4469/4740 train_time:973066ms step_avg:217.74ms
step:4470/4740 train_time:973389ms step_avg:217.76ms
step:4471/4740 train_time:973567ms step_avg:217.75ms
step:4472/4740 train_time:973885ms step_avg:217.77ms
step:4473/4740 train_time:974067ms step_avg:217.77ms
step:4474/4740 train_time:974388ms step_avg:217.79ms
step:4475/4740 train_time:974564ms step_avg:217.78ms
step:4476/4740 train_time:974885ms step_avg:217.80ms
step:4477/4740 train_time:975062ms step_avg:217.79ms
step:4478/4740 train_time:975386ms step_avg:217.82ms
step:4479/4740 train_time:975563ms step_avg:217.81ms
step:4480/4740 train_time:975887ms step_avg:217.83ms
step:4481/4740 train_time:976063ms step_avg:217.82ms
step:4482/4740 train_time:976388ms step_avg:217.85ms
step:4483/4740 train_time:976562ms step_avg:217.84ms
step:4484/4740 train_time:976885ms step_avg:217.86ms
step:4485/4740 train_time:977061ms step_avg:217.85ms
step:4486/4740 train_time:977385ms step_avg:217.87ms
step:4487/4740 train_time:977563ms step_avg:217.87ms
step:4488/4740 train_time:977885ms step_avg:217.89ms
step:4489/4740 train_time:978062ms step_avg:217.88ms
step:4490/4740 train_time:978379ms step_avg:217.90ms
step:4491/4740 train_time:978560ms step_avg:217.89ms
step:4492/4740 train_time:978882ms step_avg:217.92ms
step:4493/4740 train_time:979058ms step_avg:217.91ms
step:4494/4740 train_time:979384ms step_avg:217.93ms
step:4495/4740 train_time:979562ms step_avg:217.92ms
step:4496/4740 train_time:979885ms step_avg:217.95ms
step:4497/4740 train_time:980062ms step_avg:217.94ms
step:4498/4740 train_time:980386ms step_avg:217.96ms
step:4499/4740 train_time:980561ms step_avg:217.95ms
step:4500/4740 train_time:980884ms step_avg:217.97ms
step:4500/4740 val_loss:2.9337 train_time:981008ms step_avg:218.00ms
step:4501/4740 train_time:981064ms step_avg:217.97ms
step:4502/4740 train_time:981398ms step_avg:217.99ms
step:4503/4740 train_time:981577ms step_avg:217.98ms
step:4504/4740 train_time:981890ms step_avg:218.00ms
step:4505/4740 train_time:982067ms step_avg:217.99ms
step:4506/4740 train_time:982390ms step_avg:218.02ms
step:4507/4740 train_time:982576ms step_avg:218.01ms
step:4508/4740 train_time:982893ms step_avg:218.03ms
step:4509/4740 train_time:983074ms step_avg:218.02ms
step:4510/4740 train_time:983396ms step_avg:218.05ms
step:4511/4740 train_time:983578ms step_avg:218.04ms
step:4512/4740 train_time:983900ms step_avg:218.06ms
step:4513/4740 train_time:984077ms step_avg:218.05ms
step:4514/4740 train_time:984401ms step_avg:218.08ms
step:4515/4740 train_time:984578ms step_avg:218.07ms
step:4516/4740 train_time:984902ms step_avg:218.09ms
step:4517/4740 train_time:985080ms step_avg:218.08ms
step:4518/4740 train_time:985405ms step_avg:218.11ms
step:4519/4740 train_time:985583ms step_avg:218.10ms
step:4520/4740 train_time:985904ms step_avg:218.12ms
step:4521/4740 train_time:986081ms step_avg:218.11ms
step:4522/4740 train_time:986403ms step_avg:218.13ms
step:4523/4740 train_time:986583ms step_avg:218.13ms
step:4524/4740 train_time:986905ms step_avg:218.15ms
step:4525/4740 train_time:987084ms step_avg:218.14ms
step:4526/4740 train_time:987403ms step_avg:218.16ms
step:4527/4740 train_time:987586ms step_avg:218.15ms
step:4528/4740 train_time:987905ms step_avg:218.18ms
step:4529/4740 train_time:988084ms step_avg:218.17ms
step:4530/4740 train_time:988405ms step_avg:218.19ms
step:4531/4740 train_time:988584ms step_avg:218.18ms
step:4532/4740 train_time:988905ms step_avg:218.20ms
step:4533/4740 train_time:989085ms step_avg:218.20ms
step:4534/4740 train_time:989408ms step_avg:218.22ms
step:4535/4740 train_time:989586ms step_avg:218.21ms
step:4536/4740 train_time:989911ms step_avg:218.23ms
step:4537/4740 train_time:990088ms step_avg:218.23ms
step:4538/4740 train_time:990415ms step_avg:218.25ms
step:4539/4740 train_time:990593ms step_avg:218.24ms
step:4540/4740 train_time:990914ms step_avg:218.26ms
step:4541/4740 train_time:991093ms step_avg:218.25ms
step:4542/4740 train_time:991412ms step_avg:218.28ms
step:4543/4740 train_time:991595ms step_avg:218.27ms
step:4544/4740 train_time:991914ms step_avg:218.29ms
step:4545/4740 train_time:992091ms step_avg:218.28ms
step:4546/4740 train_time:992415ms step_avg:218.31ms
step:4547/4740 train_time:992596ms step_avg:218.30ms
step:4548/4740 train_time:992916ms step_avg:218.32ms
step:4549/4740 train_time:993098ms step_avg:218.31ms
step:4550/4740 train_time:993424ms step_avg:218.33ms
step:4551/4740 train_time:993602ms step_avg:218.33ms
step:4552/4740 train_time:993925ms step_avg:218.35ms
step:4553/4740 train_time:994103ms step_avg:218.34ms
step:4554/4740 train_time:994427ms step_avg:218.36ms
step:4555/4740 train_time:994605ms step_avg:218.35ms
step:4556/4740 train_time:994927ms step_avg:218.38ms
step:4557/4740 train_time:995106ms step_avg:218.37ms
step:4558/4740 train_time:995425ms step_avg:218.39ms
step:4559/4740 train_time:995608ms step_avg:218.38ms
step:4560/4740 train_time:995932ms step_avg:218.41ms
step:4561/4740 train_time:996112ms step_avg:218.40ms
step:4562/4740 train_time:996432ms step_avg:218.42ms
step:4563/4740 train_time:996611ms step_avg:218.41ms
step:4564/4740 train_time:996938ms step_avg:218.44ms
step:4565/4740 train_time:997116ms step_avg:218.43ms
step:4566/4740 train_time:997433ms step_avg:218.45ms
step:4567/4740 train_time:997615ms step_avg:218.44ms
step:4568/4740 train_time:997937ms step_avg:218.46ms
step:4569/4740 train_time:998118ms step_avg:218.45ms
step:4570/4740 train_time:998440ms step_avg:218.48ms
step:4571/4740 train_time:998618ms step_avg:218.47ms
step:4572/4740 train_time:998942ms step_avg:218.49ms
step:4573/4740 train_time:999120ms step_avg:218.48ms
step:4574/4740 train_time:999440ms step_avg:218.50ms
step:4575/4740 train_time:999619ms step_avg:218.50ms
step:4576/4740 train_time:999941ms step_avg:218.52ms
step:4577/4740 train_time:1000115ms step_avg:218.51ms
step:4578/4740 train_time:1000434ms step_avg:218.53ms
step:4579/4740 train_time:1000615ms step_avg:218.52ms
step:4580/4740 train_time:1000935ms step_avg:218.54ms
step:4581/4740 train_time:1001117ms step_avg:218.54ms
step:4582/4740 train_time:1001436ms step_avg:218.56ms
step:4583/4740 train_time:1001620ms step_avg:218.55ms
step:4584/4740 train_time:1001943ms step_avg:218.57ms
step:4585/4740 train_time:1002123ms step_avg:218.57ms
step:4586/4740 train_time:1002441ms step_avg:218.59ms
step:4587/4740 train_time:1002616ms step_avg:218.58ms
step:4588/4740 train_time:1002937ms step_avg:218.60ms
step:4589/4740 train_time:1003117ms step_avg:218.59ms
step:4590/4740 train_time:1003440ms step_avg:218.61ms
step:4591/4740 train_time:1003620ms step_avg:218.61ms
step:4592/4740 train_time:1003942ms step_avg:218.63ms
step:4593/4740 train_time:1004122ms step_avg:218.62ms
step:4594/4740 train_time:1004443ms step_avg:218.64ms
step:4595/4740 train_time:1004622ms step_avg:218.63ms
step:4596/4740 train_time:1004944ms step_avg:218.66ms
step:4597/4740 train_time:1005121ms step_avg:218.65ms
step:4598/4740 train_time:1005443ms step_avg:218.67ms
step:4599/4740 train_time:1005622ms step_avg:218.66ms
step:4600/4740 train_time:1005945ms step_avg:218.68ms
step:4601/4740 train_time:1006122ms step_avg:218.67ms
step:4602/4740 train_time:1006444ms step_avg:218.70ms
step:4603/4740 train_time:1006623ms step_avg:218.69ms
step:4604/4740 train_time:1006945ms step_avg:218.71ms
step:4605/4740 train_time:1007122ms step_avg:218.70ms
step:4606/4740 train_time:1007443ms step_avg:218.72ms
step:4607/4740 train_time:1007623ms step_avg:218.72ms
step:4608/4740 train_time:1007943ms step_avg:218.74ms
step:4609/4740 train_time:1008122ms step_avg:218.73ms
step:4610/4740 train_time:1008445ms step_avg:218.75ms
step:4611/4740 train_time:1008622ms step_avg:218.74ms
step:4612/4740 train_time:1008944ms step_avg:218.76ms
step:4613/4740 train_time:1009122ms step_avg:218.76ms
step:4614/4740 train_time:1009447ms step_avg:218.78ms
step:4615/4740 train_time:1009626ms step_avg:218.77ms
step:4616/4740 train_time:1009948ms step_avg:218.79ms
step:4617/4740 train_time:1010128ms step_avg:218.78ms
step:4618/4740 train_time:1010449ms step_avg:218.81ms
step:4619/4740 train_time:1010625ms step_avg:218.80ms
step:4620/4740 train_time:1010948ms step_avg:218.82ms
step:4621/4740 train_time:1011126ms step_avg:218.81ms
step:4622/4740 train_time:1011447ms step_avg:218.83ms
step:4623/4740 train_time:1011623ms step_avg:218.82ms
step:4624/4740 train_time:1011945ms step_avg:218.85ms
step:4625/4740 train_time:1012125ms step_avg:218.84ms
step:4626/4740 train_time:1012443ms step_avg:218.86ms
step:4627/4740 train_time:1012626ms step_avg:218.85ms
step:4628/4740 train_time:1012950ms step_avg:218.87ms
step:4629/4740 train_time:1013127ms step_avg:218.87ms
step:4630/4740 train_time:1013447ms step_avg:218.89ms
step:4631/4740 train_time:1013627ms step_avg:218.88ms
step:4632/4740 train_time:1013945ms step_avg:218.90ms
step:4633/4740 train_time:1014127ms step_avg:218.89ms
step:4634/4740 train_time:1014452ms step_avg:218.91ms
step:4635/4740 train_time:1014633ms step_avg:218.91ms
step:4636/4740 train_time:1014953ms step_avg:218.93ms
step:4637/4740 train_time:1015132ms step_avg:218.92ms
step:4638/4740 train_time:1015449ms step_avg:218.94ms
step:4639/4740 train_time:1015629ms step_avg:218.93ms
step:4640/4740 train_time:1015950ms step_avg:218.95ms
step:4641/4740 train_time:1016127ms step_avg:218.95ms
step:4642/4740 train_time:1016444ms step_avg:218.97ms
step:4643/4740 train_time:1016625ms step_avg:218.96ms
step:4644/4740 train_time:1016944ms step_avg:218.98ms
step:4645/4740 train_time:1017122ms step_avg:218.97ms
step:4646/4740 train_time:1017445ms step_avg:218.99ms
step:4647/4740 train_time:1017624ms step_avg:218.99ms
step:4648/4740 train_time:1017947ms step_avg:219.01ms
step:4649/4740 train_time:1018124ms step_avg:219.00ms
step:4650/4740 train_time:1018443ms step_avg:219.02ms
step:4651/4740 train_time:1018626ms step_avg:219.01ms
step:4652/4740 train_time:1018950ms step_avg:219.03ms
step:4653/4740 train_time:1019125ms step_avg:219.03ms
step:4654/4740 train_time:1019444ms step_avg:219.05ms
step:4655/4740 train_time:1019623ms step_avg:219.04ms
step:4656/4740 train_time:1019949ms step_avg:219.06ms
step:4657/4740 train_time:1020126ms step_avg:219.05ms
step:4658/4740 train_time:1020447ms step_avg:219.07ms
step:4659/4740 train_time:1020627ms step_avg:219.07ms
step:4660/4740 train_time:1020950ms step_avg:219.09ms
step:4661/4740 train_time:1021124ms step_avg:219.08ms
step:4662/4740 train_time:1021445ms step_avg:219.10ms
step:4663/4740 train_time:1021624ms step_avg:219.09ms
step:4664/4740 train_time:1021945ms step_avg:219.11ms
step:4665/4740 train_time:1022122ms step_avg:219.10ms
step:4666/4740 train_time:1022446ms step_avg:219.13ms
step:4667/4740 train_time:1022622ms step_avg:219.12ms
step:4668/4740 train_time:1022945ms step_avg:219.14ms
step:4669/4740 train_time:1023124ms step_avg:219.13ms
step:4670/4740 train_time:1023444ms step_avg:219.15ms
step:4671/4740 train_time:1023623ms step_avg:219.14ms
step:4672/4740 train_time:1023943ms step_avg:219.17ms
step:4673/4740 train_time:1024124ms step_avg:219.16ms
step:4674/4740 train_time:1024442ms step_avg:219.18ms
step:4675/4740 train_time:1024626ms step_avg:219.17ms
step:4676/4740 train_time:1024943ms step_avg:219.19ms
step:4677/4740 train_time:1025122ms step_avg:219.18ms
step:4678/4740 train_time:1025441ms step_avg:219.20ms
step:4679/4740 train_time:1025623ms step_avg:219.20ms
step:4680/4740 train_time:1025945ms step_avg:219.22ms
step:4681/4740 train_time:1026121ms step_avg:219.21ms
step:4682/4740 train_time:1026440ms step_avg:219.23ms
step:4683/4740 train_time:1026619ms step_avg:219.22ms
step:4684/4740 train_time:1026942ms step_avg:219.24ms
step:4685/4740 train_time:1027123ms step_avg:219.24ms
step:4686/4740 train_time:1027444ms step_avg:219.26ms
step:4687/4740 train_time:1027622ms step_avg:219.25ms
step:4688/4740 train_time:1027943ms step_avg:219.27ms
step:4689/4740 train_time:1028121ms step_avg:219.26ms
step:4690/4740 train_time:1028442ms step_avg:219.28ms
step:4691/4740 train_time:1028620ms step_avg:219.28ms
step:4692/4740 train_time:1028940ms step_avg:219.30ms
step:4693/4740 train_time:1029119ms step_avg:219.29ms
step:4694/4740 train_time:1029441ms step_avg:219.31ms
step:4695/4740 train_time:1029619ms step_avg:219.30ms
step:4696/4740 train_time:1029940ms step_avg:219.32ms
step:4697/4740 train_time:1030120ms step_avg:219.31ms
step:4698/4740 train_time:1030441ms step_avg:219.34ms
step:4699/4740 train_time:1030618ms step_avg:219.33ms
step:4700/4740 train_time:1030939ms step_avg:219.35ms
step:4701/4740 train_time:1031120ms step_avg:219.34ms
step:4702/4740 train_time:1031440ms step_avg:219.36ms
step:4703/4740 train_time:1031620ms step_avg:219.35ms
step:4704/4740 train_time:1031943ms step_avg:219.38ms
step:4705/4740 train_time:1032122ms step_avg:219.37ms
step:4706/4740 train_time:1032444ms step_avg:219.39ms
step:4707/4740 train_time:1032626ms step_avg:219.38ms
step:4708/4740 train_time:1032945ms step_avg:219.40ms
step:4709/4740 train_time:1033126ms step_avg:219.39ms
step:4710/4740 train_time:1033447ms step_avg:219.42ms
step:4711/4740 train_time:1033627ms step_avg:219.41ms
step:4712/4740 train_time:1033947ms step_avg:219.43ms
step:4713/4740 train_time:1034127ms step_avg:219.42ms
step:4714/4740 train_time:1034448ms step_avg:219.44ms
step:4715/4740 train_time:1034624ms step_avg:219.43ms
step:4716/4740 train_time:1034947ms step_avg:219.45ms
step:4717/4740 train_time:1035125ms step_avg:219.45ms
step:4718/4740 train_time:1035449ms step_avg:219.47ms
step:4719/4740 train_time:1035627ms step_avg:219.46ms
step:4720/4740 train_time:1035947ms step_avg:219.48ms
step:4721/4740 train_time:1036125ms step_avg:219.47ms
step:4722/4740 train_time:1036448ms step_avg:219.49ms
step:4723/4740 train_time:1036627ms step_avg:219.48ms
step:4724/4740 train_time:1036945ms step_avg:219.51ms
step:4725/4740 train_time:1037126ms step_avg:219.50ms
step:4726/4740 train_time:1037447ms step_avg:219.52ms
step:4727/4740 train_time:1037625ms step_avg:219.51ms
step:4728/4740 train_time:1037944ms step_avg:219.53ms
step:4729/4740 train_time:1038124ms step_avg:219.52ms
step:4730/4740 train_time:1038445ms step_avg:219.54ms
step:4731/4740 train_time:1038624ms step_avg:219.54ms
step:4732/4740 train_time:1038946ms step_avg:219.56ms
step:4733/4740 train_time:1039126ms step_avg:219.55ms
step:4734/4740 train_time:1039447ms step_avg:219.57ms
step:4735/4740 train_time:1039627ms step_avg:219.56ms
step:4736/4740 train_time:1039948ms step_avg:219.58ms
step:4737/4740 train_time:1040126ms step_avg:219.57ms
step:4738/4740 train_time:1040449ms step_avg:219.60ms
step:4739/4740 train_time:1040626ms step_avg:219.59ms
step:4740/4740 train_time:1040948ms step_avg:219.61ms
step:4740/4740 val_loss:2.9165 train_time:1041071ms step_avg:219.64ms
peak memory allocated: 57397 MiB reserved: 61912 MiB