Video-CoM: Interactive Video Reasoning via Chain of Manipulations
AI & ML interests
Natural Language Processing, Machine Learning, and Computer Vision
Recent Activity
View all activity
Papers
When Background Matters: Breaking Medical Vision Language Models by Transferable Attack
Counting to Four is still a Chore for VLMs
A framework for deepfake detection and reasoning supervision at scale.
Medical Foundation Model
-
MBZUAI/MedMO-8B
Image-Text-to-Text • 9B • Updated • 2.96k • 10 -
MBZUAI/MedMO-4B
Image-Text-to-Text • 4B • Updated • 2.09k • 15 -
MedMO: Grounding and Understanding Multimodal Large Language Model for Medical Images
Paper • 2602.06965 • Published • 7 -
MBZUAI/MedMO-8B-Next
Image-Text-to-Text • 9B • Updated • 4.84k • 13
Training datasets for FinMMEval Lab @CLEF'2026
Official training and dev datasets for NADI 2025 Subtask 3 (Diacritic Restoration) Shared Task
Pixel Grounding Large Multimodal Model in Remote Sensing
Open source project for Arabic Speech Recognition and Generation
Grounding Large Multimodal Model (GLaMM), the first-of-its-kind model capable of generating natural language responses that are seamlessly integrated.
Extending Visual Capabilities of LLaVA with LLaMA-3 and Phi-3
-
LLaVA++ (LLaMA-3-V)
👁33Start a chatbot server for text-based interactions
-
LLaVA++ (Phi-3-V)
👁25Launch a chatbot with image and text understanding
-
MBZUAI/LLaVA-Phi-3-mini-4k-instruct
Text Generation • 4B • Updated • 169 • 21 -
MBZUAI/LLaVA-Meta-Llama-3-8B-Instruct-FT
Text Generation • 8B • Updated • 15 • 12
Collection of MobiLlama Language Models.
Collection of ViT models trained using SatMAE++ approach.
Open Ended Medical Reinforcement Learning
The OpenEarthAgent Collection brings together the OpenEarthAgent model and its accompanying large-scale tool-augmented geospatial reasoning data.
VideoMathQA: Benchmarking Mathematical Reasoning via Multimodal Understanding in Videos
Large-scale dataset and model suite for cross-architecture GPU code transpilation between CUDA and HIP at both source and assembly levels
BiMediX2 : Bio-Medical EXpert LMM for Diverse Medical Modalities
VideoGPT+: Integrating Image and Video Encoders for Enhanced Video Understanding
"Video-ChatGPT" is a video conversation model capable of generating meaningful conversation about videos.
PALO: A Polyglot Large Multimodal Model for 5B People
GeoChat is the first grounded Large Vision Language Model, specifically tailored to Remote Sensing(RS) scenarios.
Video-CoM: Interactive Video Reasoning via Chain of Manipulations
Open Ended Medical Reinforcement Learning
A framework for deepfake detection and reasoning supervision at scale.
The OpenEarthAgent Collection brings together the OpenEarthAgent model and its accompanying large-scale tool-augmented geospatial reasoning data.
Medical Foundation Model
-
MBZUAI/MedMO-8B
Image-Text-to-Text • 9B • Updated • 2.96k • 10 -
MBZUAI/MedMO-4B
Image-Text-to-Text • 4B • Updated • 2.09k • 15 -
MedMO: Grounding and Understanding Multimodal Large Language Model for Medical Images
Paper • 2602.06965 • Published • 7 -
MBZUAI/MedMO-8B-Next
Image-Text-to-Text • 9B • Updated • 4.84k • 13
Training datasets for FinMMEval Lab @CLEF'2026
VideoMathQA: Benchmarking Mathematical Reasoning via Multimodal Understanding in Videos
Official training and dev datasets for NADI 2025 Subtask 3 (Diacritic Restoration) Shared Task
Large-scale dataset and model suite for cross-architecture GPU code transpilation between CUDA and HIP at both source and assembly levels
Pixel Grounding Large Multimodal Model in Remote Sensing
BiMediX2 : Bio-Medical EXpert LMM for Diverse Medical Modalities
Open source project for Arabic Speech Recognition and Generation
VideoGPT+: Integrating Image and Video Encoders for Enhanced Video Understanding
Grounding Large Multimodal Model (GLaMM), the first-of-its-kind model capable of generating natural language responses that are seamlessly integrated.
"Video-ChatGPT" is a video conversation model capable of generating meaningful conversation about videos.
Extending Visual Capabilities of LLaVA with LLaMA-3 and Phi-3
-
LLaVA++ (LLaMA-3-V)
👁33Start a chatbot server for text-based interactions
-
LLaVA++ (Phi-3-V)
👁25Launch a chatbot with image and text understanding
-
MBZUAI/LLaVA-Phi-3-mini-4k-instruct
Text Generation • 4B • Updated • 169 • 21 -
MBZUAI/LLaVA-Meta-Llama-3-8B-Instruct-FT
Text Generation • 8B • Updated • 15 • 12
PALO: A Polyglot Large Multimodal Model for 5B People
Collection of MobiLlama Language Models.
GeoChat is the first grounded Large Vision Language Model, specifically tailored to Remote Sensing(RS) scenarios.
Collection of ViT models trained using SatMAE++ approach.