HoneyDataV2/Honey-Data-V2
Viewer • Updated • 19.7M • 95k • 81
Bee
Note In September 2026, we released Honey-Data-V2, the corpus of the journal version of Bee (Computational Visual Media): 44.3M samples comprising 100.3M QA pairs, built by reapplying HoneyPipe to Honey-Data-15M at a larger scale.
Open-Bee/Honey-Data-15M
Viewer • Updated • 14.9M • 15.4k • 121Note On November 1, 2025, we released the Hoeny-Data-15M. This dataset was used to train the stage3 of Bee-8B.
-
Open-Bee/Bee-8B-RL
Image-Text-to-Text • 9B • Updated • 10.7k • 78 -
Open-Bee/Bee-8B-SFT
Image-Text-to-Text • 9B • Updated • 805 • 14 -
Bee: A High-Quality Corpus and Full-Stack Suite to Unlock Advanced Fully Open MLLMs
Paper • 2510.13795 • Published • 61
Open-Bee/Honey-Data-1M
Viewer • Updated • 1.07M • 542 • 22Note On November 2, 2025, we released the Hoeny-Data-1M. This dataset was used to train the stage4 of Bee-8B.
Open-Bee/Bee-Training-Data-Stage2
Viewer • Updated • 14M • 346 • 6Note On November 5, 2025, we released the Bee-Training-Data-Stage2. This dataset was used to train the stage2 of Bee-8B.
Open-Bee/Bee-Training-Data-Stage1
Viewer • Updated • 995k • 219 • 4Note On November 5, 2025, we released the Bee-Training-Data-Stage1. This dataset was used to train the stage1 of Bee-8B.
-
Open-Bee/Bee-8B-Stage1
Image-Text-to-Text • Updated • 2 -
Open-Bee/Bee-8B-Stage2
Image-Text-to-Text • 9B • Updated • 32 • 1 -
Open-Bee/Bee-8B-Stage3
Image-Text-to-Text • 9B • Updated • 36 • 1