Open-source industrial-grade ASR models supporting Mandarin, Chinese dialects and English, achieving a new SOTA on public Mandarin ASR benchmarks, while also offering outstanding singing lyrics recognition capability.

Official implementation of paper "MiniGPT-5: Interleaved Vision-and-Language Generation via Generative Vokens"

Research Code for Multimodal-Cognition Team in Ant Group

[ICCV25 Oral] Token Activation Map to Visually Explain Multimodal LLMs

Official repository for InfiGUI-G1. We introduce Adaptive Exploration Policy Optimization (AEPO) to overcome semantic alignment bottlenecks in GUI agents through efficient, guided exploration.

[IROS'25 Oral & NeurIPSw'24] Official implementation of "MineDreamer: Learning to Follow Instructions via Chain-of-Imagination for Simulated-World Control "

[ECCV 2024] Official PyTorch Implementation of "How Many Unicorns Are in This Image? A Safety Evaluation Benchmark for Vision LLMs"

The code repository for "Wings: Learning Multimodal LLMs without Text-only Forgetting" [NeurIPS 2024]

Official repository of the paper: Can ChatGPT Detect DeepFakes? A Study of Using Multimodal Large Language Models for Media Forensics

[NAACL 2025 Findings] Mitigating Hallucinations in Large Vision-Language Models via Summary-Guided Decoding

[ACL 2024] Dataset and Code of "ImplicitAVE: An Open-Source Dataset and Multimodal LLMs Benchmark for Implicit Attribute Value Extraction"

Efficient Test-Time Scaling for Small Vision-Language Models, official implementation of the paper, test-time scaling via test-time augmentation

Medical Report Generation And VQA (Adapting XrayGPT to Any Modality)

Streamlit app to chat with images using Multi-modal LLMs.

Official implementation of ICML 2025 paper "Understanding Multimodal LLMs Under Distribution Shifts: An Information-Theoretic Approach"

Q-HEART: ECG Question Answering via Knowledge-Informed Multimodal LLMs (ECAI 2025)

SpatialFusion-LM is a real-time spatial reasoning framework that combines neural depth, 3D reconstruction, and language-driven scene understanding.

LLaVA base model for use with Autodistill.

Kani extension for supporting vision-language models (VLMs). Comes with model-agnostic support for GPT-Vision and LLaVA.

A minimal, hackable Vision-Language Model built on Karpathy’s nanochat — add image understanding and multimodal chat for under $200 in compute.