[CVPR 2022 Oral & TPAMI 2024] MixFormer: End-to-End Tracking with Iterative Mixed Attention

MLLM-DataEngine: An Iterative Refinement Approach for MLLM

[CVPR'2024 Highlight] Official PyTorch implementation of the paper "VTimeLLM: Empower LLM to Grasp Video Moments".

verl: Volcano Engine Reinforcement Learning for LLMs

AllenAI's post-training codebase

SOTA Re-identification Methods and Toolbox

📚200+ Tensor/CUDA Cores Kernels, ⚡️flash-attn-mma, ⚡️hgemm with WMMA, MMA and CuTe (98%~100% TFLOPS of cuBLAS/FA2 🎉🎉).

Use PEFT or Full-parameter to finetune 450+ LLMs (Qwen2.5, InternLM3, GLM4, Llama3.3, Mistral, Yi1.5, Baichuan2, DeepSeek-R1, ...) and 150+ MLLMs (Qwen2.5-VL, Qwen2-Audio, Llama3.2-Vision, Llava, I…

A paper list of some recent works about Token Compress for Vit and VLM

✨✨Long-VITA: Scaling Large Multi-modal Models to 1 Million Tokens with Leading Short-Context Accuracy

A jounery to real multimodel R1 ! We are doing on large-scale experiment

Witness the aha moment of VLM with less than $3.

Clean, minimal, accessible reproduction of DeepSeek R1-Zero

Reproduce R1 Zero on Logic Puzzle

Fully open reproduction of DeepSeek-R1

This is a replicate of DeepSeek-R1-Zero and DeepSeek-R1 training on small models with limited data

Freeing data processing from scripting madness by providing a set of platform-agnostic customizable pipeline processing blocks.

[CVPR 2024] CapsFusion: Rethinking Image-Text Data at Scale

The implementation of "Prismer: A Vision-Language Model with Multi-Task Experts".

Caption-Anything is a versatile tool combining image segmentation, visual captioning, and ChatGPT, generating tailored captions with diverse controls for user preferences. https://huggingface.co/sp…

万卷1.0多模态语料

Paddle Multimodal Integration and eXploration, supporting mainstream multi-modal tasks, including end-to-end large-scale multi-modal pretrain models and diffusion model toolbox. Equipped with high …

Performance instrumentation and tracing for Android, Linux and Chrome (read-only mirror of https://android.googlesource.com/platform/external/perfetto/)

FlagGems is an operator library for large language models implemented in Triton Language.

DeepSeek-VL2: Mixture-of-Experts Vision-Language Models for Advanced Multimodal Understanding

[CVPR 2025] Open-source, End-to-end, Vision-Language-Action model for GUI Agent & Computer Use.

Efficient Triton Kernels for LLM Training