On-policy distillation for agents without the environment: replay teacher prefixes instead of live rollouts.
👋 Open to Work
Baohao Liao
baohao
AI & ML interests
NLP
Organizations
models 46
baohao/ReOPD_Math-Search_Qwen3-4B-Instruct-2507_SFT-RL_to_Both_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 6
baohao/OPD_Math-Search_Qwen3-4B-Instruct-2507_SFT-RL_to_Both_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 7 • 1
baohao/Both_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 7 • 1
baohao/ReOPD_Search_Qwen3-8B_SFT-RL_to_Search_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 6
baohao/OPD_Search_Qwen3-8B_SFT-RL_to_Search_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 8 • 1
baohao/ReOPD_Search_Qwen3-4B-Instruct-2507_SFT-RL_to_Search_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 9
baohao/OPD_Search_Qwen3-4B-Instruct-2507_SFT-RL_to_Search_Qwen3-4B-Instruct-2507_SFT
4B • Updated • 13 • 1
baohao/Search_Qwen3-8B_SFT-RL
8B • Updated • 7 • 1
baohao/Search_Qwen3-8B_SFT
8B • Updated • 6
baohao/Search_Qwen3-4B-Instruct-2507_SFT-RL
4B • Updated • 10 • 1
datasets 39
baohao/both-cold-start
Viewer • Updated • 4k • 20
baohao/Search_Qwen3-8B_SFT-RL_Prefix
Viewer • Updated • 141k • 11 • 1
baohao/Search_Qwen3-4B-Instruct-2507_SFT-RL_Prefix
Viewer • Updated • 145k • 12 • 2
baohao/Math_Qwen3-30B-A3B-Instruct-2507_SFT-RL_Prefix
Viewer • Updated • 54.7k • 39 • 1
baohao/Math_Qwen3-8B_SFT-RL_Prefix
Viewer • Updated • 53.8k • 28
baohao/Math_Qwen3-4B-Instruct-2507_SFT-RL_Prefix
Viewer • Updated • 81.9k • 49 • 1
baohao/math-train
Viewer • Updated • 6.4k • 32
baohao/math-cold-start
Viewer • Updated • 2k • 37 • 1
baohao/search-cold-start
Viewer • Updated • 2k • 14 • 1
baohao/search-train
Viewer • Updated • 6.5k • 28 • 1