Marc Sun's picture

Building on HF

Marc Sun

marcsun13

huggingface

·

AI & ML interests

LLM, Quantization, Training, Inference

Recent Activity

upvoted an article about 18 hours ago

Continuous batching from first principles

upvoted an article about 18 hours ago

Diffusers welcomes FLUX-2

upvoted an article about 18 hours ago

We Got Claude to Fine-Tune an Open Source LLM

View all activity

Organizations

upvoted 8 articles about 18 hours ago

Article

Continuous batching from first principles

+1

Nov 25, 2025

•

297

Article

Diffusers welcomes FLUX-2

+6

Nov 25, 2025

•

168

Article

We Got Claude to Fine-Tune an Open Source LLM

Dec 4, 2025

•

564

Article

New in llama.cpp: Model Management

27 days ago

•

104

Article

Tokenization in Transformers v5: Simpler, Clearer, and More Modular

+4

20 days ago

•

104

Article

NVIDIA brings agents to life with DGX Spark and Reachy Mini

+1

2 days ago

•

35

Article

Introducing Falcon-H1-Arabic: Pushing the Boundaries of Arabic Language AI with Hybrid Architecture

2 days ago

•

31

Article

NVIDIA Cosmos Reason 2 Brings Advanced Reasoning To Physical AI

1 day ago

•

22

liked a Space about 18 hours ago

Reachy Mini

All about Reachy Mini, from building to getting started

upvoted an article about 1 month ago

Article

Transformers v5: Simple model definitions powering the AI ecosystem

+2

Dec 1, 2025

•

265

upvoted a paper about 2 months ago

SINQ: Sinkhorn-Normalized Quantization for Calibration-Free Low-Precision LLM Weights

Paper • 2509.22944 • Published Sep 26, 2025 • 79

upvoted 2 articles 2 months ago

Article

Classement compar:IA : des votes des utilisateurs au classement participatif des modèles

Nov 3, 2025

•

6

Article

On the Shifting Global Compute Landscape

Oct 29, 2025

•

55

New activity in kernels-community/triton_kernels 4 months ago

Implement make_opt_flags function for XPU

#5 opened 4 months ago by

reacted to lysandre's post with 🚀 4 months ago

Post

7468

We're kick-starting the process of Transformers v5, with @ArthurZ and @cyrilvallez !

v5 should be significant: we're using it as a milestone for performance optimizations, saner defaults, and a much cleaner code base worthy of 2025.

Fun fact: v4.0.0-rc-1 came out on Nov 19, 2020, nearly five years ago!

6 replies

·

New activity in kernels-community/triton_kernels 4 months ago

Kernel assertion errors on 5090 using generation with MXfp4 (gpt-oss) - (stable on 4090)

#1 opened 5 months ago by

Update torch-ext/triton_kernels/matmul_ogs_details/opt_flags.py

#3 opened 4 months ago by

Update build/torch-universal/triton_kernels/matmul_ogs_details/opt_flags.py

#2 opened 4 months ago by

liked a model 4 months ago

OpenGVLab/InternVL3_5-GPT-OSS-20B-A4B-Preview

Image-Text-to-Text • 0.4B • Updated Aug 29, 2025 • 34.9k • 82

updated a model 4 months ago

kernels-community/triton_kernels

Updated Oct 27, 2025 • 2.49k • 7