Duplicate from perplexity-ai/pplx-embed-v1-4b

Co-authored-by: Bo Wang <bowang0911@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,44 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+assets/diag.png filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_1 filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_2 filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_3 filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_4 filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_5 filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_6 filter=lfs diff=lfs merge=lfs -text
+onnx/model.onnx_data_7 filter=lfs diff=lfs merge=lfs -text

1_Pooling/config.json

@@ -0,0 +1,9 @@
+{
+  "word_embedding_dimension": 2560,
+  "pooling_mode_cls_token": false,
+  "pooling_mode_mean_tokens": true,
+  "pooling_mode_max_tokens": false,
+  "pooling_mode_mean_sqrt_len_tokens": false,
+  "pooling_mode_weightedmean_tokens": false,
+  "pooling_mode_lasttoken": false
+}

README.md

@@ -0,0 +1,192 @@
+---
+license: mit
+pipeline_tag: feature-extraction
+tags:
+- feature-extraction
+- sentence-similarity
+- mteb
+- sentence-transformers
+language:
+- multilingual
+---
+
+
+<p align="center">
+  <img src="assets/logo.svg" alt="Perplexity Logo" width="400">
+</p>
+
+<p align="center">pplx-embed-v1: Diffusion-Pretrained Dense and Contextual Embeddings</p>
+
+`pplx-embed-v1` and `pplx-embed-context-v1` are state-of-the-art text embedding models optimized for real-world, web-scale retrieval tasks.
+
+- Use **`pplx-embed-v1`** for independent text embedding (queries, documents, semantic search)
+- Use **`pplx-embed-context-v1`** for document chunks in RAG systems where surrounding context matters
+
+> [!IMPORTANT]
+> `pplx-embed-v1` and `pplx-embed-context-v1` natively produce *unnormalized* int8-quantized embeddings. Ensure that you compare them via *cosine similarity*.
+
+
+![diag.png](assets/diag.png)
+
+## Models
+
+| Model | Dimensions | Context | MRL | Quantization | Instruction | Pooling |
+|:-----:|:----------:|:-------:|:---:|:------------:|:-----------:|:-------:|
+| `pplx-embed-v1-0.6B` | 1024 | 32K | Yes | INT8/BINARY | No | Mean |
+| `pplx-embed-v1-4B` | 2560 | 32K | Yes | INT8/BINARY | No | Mean |
+| `pplx-embed-context-v1-0.6B` | 1024 | 32K | Yes | INT8/BINARY | No | Mean |
+| `pplx-embed-context-v1-4B` | 2560 | 32K | Yes | INT8/BINARY | No | Mean |
+
+<sub>All models are built on diffusion continued pre-trained Qwen3 at Perplexity AI.</sub>
+
+<sub>Many modern embedding models rely on instruction tuning, where users prepend an instruction string to the text being embedded. This can yield a 2%-3% lift on benchmarks, but it also introduces prompt-selection overhead and can make indexing pipelines brittle (small instruction changes can shift embedding space). We deliberately **avoid** this requirement: you can embed the text you want to index directly, without having to choose or maintain an instruction prefix.</sub>
+
+## Usage
+
+<details>
+<summary>Via API</summary>
+
+```bash
+curl -X POST https://api.perplexity.ai/v1/embeddings \
+  -H "Authorization: Bearer YOUR_API_KEY" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "input": [
+      "Scientists explore the universe driven by curiosity.",
+      "Children learn through curious exploration.",
+      "Historical discoveries began with curious questions.",
+      "Animals use curiosity to adapt and survive.",
+      "Philosophy examines the nature of curiosity."
+    ],
+    "model": "pplx-embed-v1-4b"
+  }'
+```
+
+</details>
+
+
+<details>
+<summary>Using SentenceTransformers</summary>
+
+```python
+from sentence_transformers import SentenceTransformer
+
+model = SentenceTransformer(
+    "perplexity-ai/pplx-embed-v1-4B",
+    trust_remote_code=True
+)
+
+texts = [
+    "Scientists explore the universe driven by curiosity.",
+    "Children learn through curious exploration.",
+    "Historical discoveries began with curious questions.",
+    "Animals use curiosity to adapt and survive.",
+    "Philosophy examines the nature of curiosity.",
+]
+
+embeddings = model.encode(texts) # Shape: (5, 2560), quantized to int8
+embeddings = model.encode(texts, quantization="binary") # Shape: (5, 2560), quantized to binary
+```
+
+</details>
+
+<details>
+<summary> Using ONNX models </summary>
+
+```python
+
+import onnxruntime as ort
+from transformers import AutoTokenizer
+import numpy as np
+
+tokenizer = AutoTokenizer.from_pretrained("perplexity-ai/pplx-embed-v1-4b", trust_remote_code=True)
+session = ort.InferenceSession("onnx/model.onnx")
+
+
+texts = [
+    "Scientists explore the universe driven by curiosity.",
+    "Children learn through curious exploration.",
+    "Historical discoveries began with curious questions.",
+    "Animals use curiosity to adapt and survive.",
+    "Philosophy examines the nature of curiosity.",
+]
+
+tokenized = tokenizer(
+    texts,
+    padding=True,
+    truncation=True,
+    return_tensors="np"
+)
+
+onnx_inputs = {
+    "input_ids": tokenized["input_ids"].astype(np.int64),
+    "attention_mask": tokenized["attention_mask"].astype(np.int64),
+}
+
+# Run inference
+onnx_embeddings = session.run([out.name for out in session.get_outputs()], onnx_inputs)
+
+# ONNX produces both int8 and binary precision embeddings:
+int8_embeddings = onnx_embeddings[2]
+binary_embeddings = onnx_embeddings[3]
+packed_embeddings = np.packbits(binary_embeddings != -1, axis=-1)
+```
+
+</details>
+
+<details>
+<summary>Using Text Embeddings Inference (TEI)</summary>
+
+> [!NOTE]
+> Text Embeddings Inference v1.9.2+ is required.
+
+> [!IMPORTANT]
+> Currently, only int8-quantized embeddings are available via TEI. Remember to use cosine similarity with unnormalized int8 embeddings.
+
+- CPU w/ Candle:
+
+```bash
+docker run -p 8080:80 ghcr.io/huggingface/text-embeddings-inference:cpu-1.9 --model-id perplexity-ai/pplx-embed-v1-4B --dtype float32
+```
+
+- CPU w/ ORT (ONNX Runtime):
+
+```bash
+docker run -p 8080:80 ghcr.io/huggingface/text-embeddings-inference:cpu-1.9 --model-id onnx-community/pplx-embed-v1-4B --dtype float32
+```
+
+- GPU w/ CUDA:
+
+```bash
+docker run --gpus all --shm-size 1g -p 8080:80 ghcr.io/huggingface/text-embeddings-inference:cuda-1.9 --model-id perplexity-ai/pplx-embed-v1-4B --dtype float32
+```
+
+> If you hit OOM during warmup, lower --max-batch-tokens and --max-client-batch-size. Set --max-batch-tokens to max_sequence_length × batch_size (e.g., 2048 tokens × 8 sequences = 16384).
+
+> Alternatively, when running in CUDA you can use the architecture / compute capability specific
+> container instead of the `cuda-1.9`, as that includes the binaries for Turing, Ampere, Hopper and
+> Blackwell, so using a dedicated container will be lighter e.g., `ampere-1.9`.
+
+And then you can send requests to it via cURL to `/embed`:
+
+```bash
+curl http://0.0.0.0:8080/embed \
+  -H "Content-Type: application/json" \
+  -d '{
+    "inputs": [
+      "Scientists explore the universe driven by curiosity.",
+      "Children learn through curious exploration.",
+      "Historical discoveries began with curious questions.",
+      "Animals use curiosity to adapt and survive.",
+      "Philosophy examines the nature of curiosity."
+    ],
+    "normalize": false
+  }'
+```
+
+</details>
+
+## Technical Details
+
+For comprehensive technical details and evaluation results, see our paper on arXiv: https://arxiv.org/abs/2602.11151.
+

added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

config.json

@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "PPLXQwen3Model"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration.PPLXQwen3Config",
+    "AutoModel": "modeling.PPLXQwen3Model"
+  },
+  "bos_token_id": 151643,
+  "dtype": "float32",
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 2560,
+  "initializer_range": 0.02,
+  "intermediate_size": 9728,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 36,
+  "model_type": "bidirectional_pplx_qwen3",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 36,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_parameters": {
+    "rope_theta": 1000000,
+    "rope_type": "default"
+  },
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.0.0.dev0",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936,
+  "attn_implementation": "sdpa",
+  "use_bidirectional_attention": true
+}

configuration.py

@@ -0,0 +1,5 @@
+from transformers.models.qwen3.configuration_qwen3 import Qwen3Config
+
+
+class PPLXQwen3Config(Qwen3Config):
+    model_type = "bidirectional_pplx_qwen3"

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35f5ccd0a3c772e1a580d10056c294a960e5f2ee7c12e6997171541fbabf4a09
+size 16089915848

modeling.py

@@ -0,0 +1,83 @@
+from typing import Callable
+import torch
+from transformers import Qwen3Model
+from transformers.cache_utils import Cache
+from transformers.masking_utils import create_causal_mask
+from transformers.modeling_outputs import BaseModelOutputWithPooling
+from transformers.processing_utils import Unpack
+from transformers.utils import TransformersKwargs
+from .configuration import PPLXQwen3Config
+
+
+# From modeling_t5gemma.py
+def bidirectional_mask_function(attention_mask: torch.Tensor | None) -> Callable:
+    """
+    This creates bidirectional attention mask.
+    """
+
+    def inner_mask(batch_idx: int, head_idx: int, q_idx: int, kv_idx: int) -> bool:
+        if attention_mask is None:
+            return torch.ones((), dtype=torch.bool)
+        return attention_mask[batch_idx, kv_idx].to(torch.bool)
+
+    return inner_mask
+
+
+class PPLXQwen3Model(Qwen3Model):
+    _supports_flash_attn = True
+    _supports_sdpa = True
+
+    config_class = PPLXQwen3Config
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.post_init()
+
+    def post_init(self):
+        super().post_init()
+        # Override to set all layers to non-causal attention. This'll work with attn_implementation="flash_attention_2" or "sdpa"
+        for layer in self.layers:
+            layer.self_attn.is_causal = False
+
+    def forward(
+        self,
+        input_ids: torch.LongTensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.LongTensor | None = None,
+        past_key_values: Cache | None = None,
+        inputs_embeds: torch.FloatTensor | None = None,
+        use_cache: bool | None = None,
+        cache_position: torch.LongTensor | None = None,
+        **kwargs: Unpack[TransformersKwargs],
+    ) -> BaseModelOutputWithPooling:
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+            input_ids = None
+
+        # We construct a dummy tensor imitating initial positions
+        dummy_cache_position = torch.arange(
+            inputs_embeds.shape[1], device=inputs_embeds.device, dtype=torch.long
+        )
+        attention_mask = {
+            "full_attention": create_causal_mask(
+                config=self.config,
+                input_embeds=inputs_embeds,
+                attention_mask=attention_mask,
+                cache_position=dummy_cache_position,
+                past_key_values=None,
+                position_ids=position_ids,
+                or_mask_function=bidirectional_mask_function(attention_mask),
+            )
+        }
+
+        outputs = super().forward(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            cache_position=cache_position,
+            **kwargs,
+        )
+        return outputs

modules.json

@@ -0,0 +1,21 @@
+[
+  {
+    "idx": 0,
+    "name": "0",
+    "path": "",
+    "type": "sentence_transformers.models.Transformer"
+  },
+  {
+    "idx": 1,
+    "name": "1",
+    "path": "1_Pooling",
+    "type": "sentence_transformers.models.Pooling"
+  },
+  {
+    "idx": 2,
+    "name": "2",
+    "path": "",
+    "type": "st_quantize.FlexibleQuantizer",
+    "kwargs": ["quantization"]
+  }
+]

onnx/model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ad2fb365d833ff0f08109bb8babaff96987bc40d22d8afa55dd9733078d2414
+size 667659

onnx/model.onnx_data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:456f8f7269b87a1774ad05734c5437708d9bce8526a2bea2af3b4035499ae2bc
+size 2081202176

onnx/model.onnx_data_1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6055f3ec48c189b140e435014ea7a964c472a2a55d4d54fa2688700743c0111a
+size 2018611200

onnx/model.onnx_data_2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bfab8bf7dee4b9f4a154266b4cb2ae7b7af953345fbfdebec4b4d1312d3d3e4
+size 2018611200

onnx/model.onnx_data_3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5b008d0fb9797fe5fcc4f3b2f7a5e598c1af71d3f13556fb722578d60e9e00f
+size 2018611200

onnx/model.onnx_data_4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5430d76ddafbdc644b6a5338c8dd01577644a54f8efb1312f921843b8e6d1871
+size 2018611200

onnx/model.onnx_data_5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c66ace38112b8c1bea0643c8491a106f40ce171b6ecfcdaac6447484676b3854
+size 2018611200

onnx/model.onnx_data_6

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37642276adabd79d9c9d9c5b052e4a7974e95d634b1a5c47a4cbb4d07a3d0245
+size 2018611200

onnx/model.onnx_data_7

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d452f4b217363fe0c641f522c03b1129e2077fd371e3040026ade2de87cc4a86
+size 1913743360

special_tokens_map.json

@@ -0,0 +1,39 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "mask_token": {
+    "content": "â½Ĺ",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": "<|endoftext|>"
+}

st_quantize.py

@@ -0,0 +1,122 @@
+import torch
+import numpy as np
+from typing import Literal
+from sentence_transformers.models import Module
+
+
+class Quantizer(torch.nn.Module):
+    def __init__(self, hard: bool = True):
+        """
+        Args:
+            hard: Whether to use hard or soft quantization. Defaults to True.
+        """
+        super().__init__()
+        self._hard = hard
+
+    def _hard_quantize(self, x, *args, **kwargs) -> torch.Tensor:
+        raise NotImplementedError
+
+    def _soft_quantize(self, x, *args, **kwargs) -> torch.Tensor:
+        raise NotImplementedError
+
+    def forward(self, x, *args, **kwargs) -> torch.Tensor:
+        soft = self._soft_quantize(x, *args, **kwargs)
+
+        if not self._hard:
+            result = soft
+        else:
+            result = (
+                self._hard_quantize(x, *args, **kwargs).detach() + soft - soft.detach()
+            )
+
+        return result
+
+
+class Int8TanhQuantizer(Quantizer):
+    def __init__(
+        self,
+        hard: bool = True,
+    ):
+        super().__init__(hard=hard)
+        self.qmin = -128
+        self.qmax = 127
+
+    def _soft_quantize(self, x, *args, **kwargs):
+        return torch.tanh(x)
+
+    def _hard_quantize(self, x, *args, **kwargs):
+        soft = self._soft_quantize(x)
+        int_x = torch.round(soft * self.qmax)
+        int_x = torch.clamp(int_x, self.qmin, self.qmax)
+        return int_x
+
+
+class BinaryTanhQuantizer(Quantizer):
+    def __init__(
+        self,
+        hard: bool = True,
+        scale: float = 1.0,
+    ):
+        super().__init__(hard)
+        self._scale = scale
+
+    def _soft_quantize(self, x, *args, **kwargs):
+        return torch.tanh(self._scale * x)
+
+    def _hard_quantize(self, x, *args, **kwargs):
+        return torch.where(x >= 0, 1.0, -1.0)
+    
+
+class PackedBinaryQuantizer:
+    def __call__(self, x: torch.Tensor) -> torch.Tensor:
+        bits = np.where(x.cpu().numpy() >=  0, True, False)
+        packed = np.packbits(bits, axis=-1)
+        return torch.from_numpy(packed).to(x.device)
+
+
+class FlexibleQuantizer(Module):
+    def __init__(self):
+        super().__init__()
+        self._int8_quantizer = Int8TanhQuantizer()
+        self._binary_quantizer = BinaryTanhQuantizer()
+        self._packed_binary_quantizer = PackedBinaryQuantizer()
+
+    def forward(
+        self,
+        features: dict[str, torch.Tensor],
+        quantization: Literal["int8", "binary", "ubinary"] = "int8",
+        **kwargs
+    ) -> dict[str, torch.Tensor]:
+        if quantization == "int8":
+            features["sentence_embedding"] = self._int8_quantizer(
+                features["sentence_embedding"]
+            )
+        elif quantization == "binary":
+            features["sentence_embedding"] = self._binary_quantizer(
+                features["sentence_embedding"]
+            )
+        elif quantization == "ubinary":
+            features["sentence_embedding"] = self._packed_binary_quantizer(
+                features["sentence_embedding"]
+            )
+        else:
+            raise ValueError(
+                f"Invalid quantization type: {quantization}. Must be 'binary', 'ubinary', or 'int8'."
+            )
+        return features
+
+    @classmethod
+    def load(
+        cls,
+        model_name_or_path: str,
+        subfolder: str = "",
+        token: bool | str | None = None,
+        cache_folder: str | None = None,
+        revision: str | None = None,
+        local_files_only: bool = False,
+        **kwargs,
+    ):
+        return cls()
+        
+    def save(self, output_path: str, *args, **kwargs) -> None: 
+        return

tokenizer_config.json

@@ -0,0 +1,249 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151642": {
+      "content": "â½Ĺ",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "mask_token": "â½Ĺ",
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "sep_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}