Tony Wu

tonywu71

21 45 66

AI & ML interests

LLM, Multimodal, Agents, Information Retrieval, RAG, Speech

Recent Activity

upvoted an article 5 days ago

Lattice: an 8 MB static retriever that embeds Wikipedia in 7 minutes

upvoted an article 10 days ago

mDenseOn with the mLateOn: Open Multilingual, Long-Context, and Code Retrieval Models

upvoted an article 23 days ago

One Adapter, Both Modalities: Field Notes from Building and Serving a Multimodal Reranker

View all activity

Organizations

upvoted an article 5 days ago

Article

Lattice: an 8 MB static retriever that embeds Wikipedia in 7 minutes

erikkaum

•

5 days ago

• 19

upvoted an article 10 days ago

Article

mDenseOn with the mLateOn: Open Multilingual, Long-Context, and Code Retrieval Models

lightonai

•

13 days ago

• 35

upvoted an article 23 days ago

Article

One Adapter, Both Modalities: Field Notes from Building and Serving a Multimodal Reranker

lightonai

•

27 days ago

• 19

liked a model about 2 months ago

theforecastingcompany/t0-alpha

Time Series Forecasting • 0.1B • Updated Jun 15 • 6.36k • 36

upvoted a collection about 2 months ago

SuperBPE

Collection

SuperBPE tokenizers and models trained with them • 8 items • Updated Mar 23 • 18

liked 2 Spaces about 2 months ago

Gemma 4 WebGPU Kernels

⚡

320

Chat with Gemma 4 E2B AI model

Encoder-Free VLM

👁

Train Your Own Encoder-Free VLM in $100

upvoted 2 articles 2 months ago

Article

Harness, Scaffold, and the AI Agent Terms Worth Getting Right

sergiopaniego, ariG23498

•

May 25

• 137

Article

Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler

ariG23498, sayakpaul, sergiopaniego, ror, pcuenq

•

May 29

• 158

liked a Space 3 months ago

The ultimate guide to RL environments: building and scaling them in the LLM era

📝

208

Building and scaling RL environments for LLM training

upvoted an article 3 months ago

Article

Unlocking asynchronicity in continuous batching

ror, pcuenq, ariG23498

•

May 14

• 65

upvoted a paper 4 months ago

Working Notes on Late Interaction Dynamics: Analyzing Targeted Behaviors of Late Interaction Models

Paper • 2603.26259 • Published Mar 27 • 8

upvoted an article 4 months ago

Article

DeepSeek-V4: a million-token context that agents can actually use

burtenshaw

•

Apr 24

• 53

liked a Space 5 months ago

The Synthetic Data Playbook: Generating Trillions of the Finest Tokens

📝

270

Visualize synthetic‑data experiments as an interactive bookshelf

upvoted 2 articles 6 months ago

Article

Mixture of Experts (MoEs) in Transformers

ariG23498, pcuenq, merve, IlyasMoutawwakil, ArthurZ, sergiopaniego, Molbap

•

Feb 26

• 175

Article

LateOn-Code & ColGrep: LightOn unveils state-of-the-art code retrieval models and code search tooling

lightonai

•

Feb 12

• 58

liked a Space 8 months ago

Evaluation Guidebook

📝

348

Explore LLM benchmark scores over time

upvoted an article 8 months ago

Article

Transformers v5: Simple model definitions powering the AI ecosystem

lysandre, ArthurZ, cyrilvallez, reach-vb

•

Dec 1, 2025

• 313

upvoted an article 9 months ago

Article

Continuous batching from first principles

ror, ArthurZ, mcpotato

•

Nov 25, 2025

• 433

authored a paper 9 months ago

Surfer 2: The Next Generation of Cross-Platform Computer Use Agents

Paper • 2510.19949 • Published Oct 22, 2025 • 38

Tony Wu

AI & ML interests

Recent Activity

Organizations

tonywu71's activity

Lattice: an 8 MB static retriever that embeds Wikipedia in 7 minutes

mDenseOn with the mLateOn: Open Multilingual, Long-Context, and Code Retrieval Models

One Adapter, Both Modalities: Field Notes from Building and Serving a Multimodal Reranker

Gemma 4 WebGPU Kernels

Encoder-Free VLM

Harness, Scaffold, and the AI Agent Terms Worth Getting Right

Profiling in PyTorch (Part 1): A Beginner's Guide to torch.profiler

The ultimate guide to RL environments: building and scaling them in the LLM era

Unlocking asynchronicity in continuous batching

DeepSeek-V4: a million-token context that agents can actually use

The Synthetic Data Playbook: Generating Trillions of the Finest Tokens

Mixture of Experts (MoEs) in Transformers

LateOn-Code & ColGrep: LightOn unveils state-of-the-art code retrieval models and code search tooling

Evaluation Guidebook

Transformers v5: Simple model definitions powering the AI ecosystem

Continuous batching from first principles