Qwen2.5 models (0.5B/1.5B/3B) from 'From Atomic to Composite: RL Enables Generalization in Complementary Reasoning'.
-
sitao/From_atomic_to_composite
Preview • Updated • 65 • 2 -
sitao/atomic-to-composite-qwen2.5-0.5b-sft-atomic
Text Generation • 0.5B • Updated • 154 -
sitao/atomic-to-composite-qwen2.5-0.5b-sft-composite
Text Generation • 0.5B • Updated • 146 -
sitao/atomic-to-composite-qwen2.5-0.5b-composite-first-rl
Text Generation • 0.6B • Updated • 155