MMVistaReason: Toward Open-data and post-training recipes for multimodal reasoning
AI & ML interests
Data-centric AI, LLM, MLLM
Recent Activity
Papers
MMVistaReason: Toward Open-Data and Post-Training Recipes for Multimodal Reasoning
Tracing the Roots: A Multi-Agent Framework for Uncovering Data Lineage in Post-Training LLMs
High-quality dataset for financial LLM post-training.
High-quality mixture datasets for post-training covering multiple domains.
-
OpenDataArena/ODA-Mixture-500k
Viewer β’ Updated β’ 506k β’ 465 β’ 124 -
OpenDataArena/ODA-Mixture-100k
Viewer β’ Updated β’ 101k β’ 138 β’ 99 -
OpenDataArena/Qwen2.5-7B-ODA-Mixture-500k
Text Generation β’ 333k β’ Updated β’ 21 β’ 2 -
OpenDataArena/Qwen2.5-7B-ODA-Mixture-100k
Text Generation β’ 333k β’ Updated β’ 15
High-quality STEM reasoning dataset for Multimodal LLM post-training.
-
OpenDataArena/MMFineReason-Full-2.3M-Qwen3-VL-235B-Thinking
Viewer β’ Updated β’ 2.29M β’ 3.03k β’ 65 -
MMFineReason: Closing the Multimodal Reasoning Gap via Open Data-Centric Methods
Paper β’ 2601.21821 β’ Published β’ 62 -
OpenDataArena/MMFineReason-SFT-123K-Qwen3-VL-235B-Thinking
Viewer β’ Updated β’ 123k β’ 733 β’ 86 -
OpenDataArena/MMFineReason-4B
Visual Question Answering β’ 5B β’ Updated β’ 37 β’ 15
ODA-Scored-Data by implemented multiple data scores.
High-quality mathematical datasets for post training.
-
OpenDataArena/MathLake
Viewer β’ Updated β’ 8.31M β’ 634 β’ 21 -
OpenDataArena/ODA-Math-460k
Viewer β’ Updated β’ 460k β’ 481 β’ 106 -
OpenDataArena/Qwen2.5-7B-ODA-Math-460k
Text Generation β’ 8B β’ Updated β’ 22 -
OpenDataArena/Qwen3-8B-ODA-Math-460k
Text Generation β’ 308k β’ Updated β’ 24 β’ 2
MMVistaReason: Toward Open-data and post-training recipes for multimodal reasoning
High-quality STEM reasoning dataset for Multimodal LLM post-training.
-
OpenDataArena/MMFineReason-Full-2.3M-Qwen3-VL-235B-Thinking
Viewer β’ Updated β’ 2.29M β’ 3.03k β’ 65 -
MMFineReason: Closing the Multimodal Reasoning Gap via Open Data-Centric Methods
Paper β’ 2601.21821 β’ Published β’ 62 -
OpenDataArena/MMFineReason-SFT-123K-Qwen3-VL-235B-Thinking
Viewer β’ Updated β’ 123k β’ 733 β’ 86 -
OpenDataArena/MMFineReason-4B
Visual Question Answering β’ 5B β’ Updated β’ 37 β’ 15
High-quality dataset for financial LLM post-training.
ODA-Scored-Data by implemented multiple data scores.
High-quality mixture datasets for post-training covering multiple domains.
-
OpenDataArena/ODA-Mixture-500k
Viewer β’ Updated β’ 506k β’ 465 β’ 124 -
OpenDataArena/ODA-Mixture-100k
Viewer β’ Updated β’ 101k β’ 138 β’ 99 -
OpenDataArena/Qwen2.5-7B-ODA-Mixture-500k
Text Generation β’ 333k β’ Updated β’ 21 β’ 2 -
OpenDataArena/Qwen2.5-7B-ODA-Mixture-100k
Text Generation β’ 333k β’ Updated β’ 15
High-quality mathematical datasets for post training.
-
OpenDataArena/MathLake
Viewer β’ Updated β’ 8.31M β’ 634 β’ 21 -
OpenDataArena/ODA-Math-460k
Viewer β’ Updated β’ 460k β’ 481 β’ 106 -
OpenDataArena/Qwen2.5-7B-ODA-Math-460k
Text Generation β’ 8B β’ Updated β’ 22 -
OpenDataArena/Qwen3-8B-ODA-Math-460k
Text Generation β’ 308k β’ Updated β’ 24 β’ 2