Ross Wightman's picture

Ross Wightman

rwightman

·

AI & ML interests

Computer vision, transfer learning, semi/self supervised learning, robotics.

Recent Activity

new activity about 21 hours ago

timm/vit_little_patch16_reg4_gap_256.sbb_in1k:Loss exploding to nan

liked a model 3 days ago

openai/gpt-oss-20b

liked a model 3 days ago

openai/gpt-oss-120b

View all activity

Organizations

upvoted a collection 3 days ago

gpt-oss

Open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases. • 2 items • Updated 1 day ago • 245

upvoted an article 4 days ago

Article

An Analysis of Chinese LLM Censorship and Bias with Qwen 2 Instruct

By

•

Jun 11, 2024

• 64

upvoted a collection 7 days ago

MetaCLIP

MetaCLIP & MetaCLIP2 OpenCLIP and timm models. All models are dual timm + OpenCLIP (or just timm for specific vit encoders). • 24 items • Updated 7 days ago • 2

upvoted a collection 14 days ago

Perception Encoder

OpenCLIP (PE Core image + text) and timm PE Core, Spatial, Lang (ViT only) weights. NOTE: These weights do not work with original modeling code. • 19 items • Updated 7 days ago • 5

upvoted 2 collections 16 days ago

Meta CLIP

Scaling CLIP data with transparent training distribution from an end-to-end pipeline. • 7 items • Updated 18 days ago • 4

Perception Encoder

17 items • Updated 28 days ago • 62

upvoted a paper about 1 month ago

RecConv: Efficient Recursive Convolutions for Multi-Frequency Representations

Paper • 2412.19628 • Published Dec 27, 2024 • 2

upvoted 2 collections about 1 month ago

RecNeXt

37 items • Updated 7 days ago • 2

Gemma 3n

4 items • Updated 29 days ago • 202

upvoted 2 collections 3 months ago

MedGemma Release

Collection of Gemma 3 variants for performance on medical text and image comprehension to accelerate building healthcare-based AI applications. • 7 items • Updated 28 days ago • 272

OpenVision

27 items • Updated May 8 • 29

upvoted a paper 6 months ago

SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features

Paper • 2502.14786 • Published Feb 20 • 146

upvoted a collection 6 months ago

SigLIP 2

OpenCLIP and timm SigLIP 2 models • 47 items • Updated 3 days ago • 23

upvoted 3 articles 6 months ago

Article

SigLIP 2: A better multilingual vision language encoder

By

and 2 others •

Feb 21

• 175

Article

🚀 Deploying OLMo-7B with Text Generation Inference (TGI) on Hugging Face Spaces

By

•

Feb 2

• 6

Article

Open-R1: a fully open reproduction of DeepSeek-R1

By

and 2 others •

Jan 28

• 876

upvoted 2 articles 7 months ago

Article

MiniMax-01 is Now Open-Source: Scaling Lightning Attention for the AI Agent Era

By

•

Jan 15

• 47

Article

Timm ❤️ Transformers: Use any timm model with transformers

By

and 4 others •

Jan 16

• 51

upvoted 2 papers 8 months ago

Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference

Paper • 2412.13663 • Published Dec 18, 2024 • 153

Phi-4 Technical Report

Paper • 2412.08905 • Published Dec 12, 2024 • 121