Repo for paper Rethinking Generalization in Reasoning SFT: A Conditional Analysis on Optimization, Data, and Model Capability.
Qihan Ren
jasonrqh
AI & ML interests
XAI, LLM reasoning & safety, Coding agent
Recent Activity
upvoted a paper 14 days ago
TraceForge: Distilling with Your Own Thinking upvoted a paper 16 days ago
BlenderWorld: Benchmarking Visual Agents for 3D Creation from Real-World References upvoted a paper 20 days ago
UOPD: Uncertainty-Aware Intervention for On-Policy Distillation of Multi-Turn AgentsOrganizations
models 33
jasonrqh/InternLM2.5-20B_DeepSeek-R1-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/InternLM2.5-20B_Numina-Math-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/InternLM2.5-20B_Math-NoCoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/InternLM2.5-20B_Countdown-CoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/InternLM2.5-20B_Math-CoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/Qwen2.5-14B_Math-CoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/Qwen2.5-7B_Math-CoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/Qwen2.5-3B_Math-CoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/Qwen2.5-1.5B_Math-CoT-20k_lr5e-5_ep8_bs256
Text Generation • Updated
jasonrqh/Qwen3-14B_Math-CoT-2.5k_lr5e-5_ep8_bs32
Text Generation • Updated
datasets 6
jasonrqh/Math-CoT-44k-Qwen3-32b-n32-16384-with-logprob-and-entropy
Viewer • Updated • 44.4k • 2.16k • 1
jasonrqh/DeepSeek-R1-20k
Viewer • Updated • 20.5k • 36
jasonrqh/NuminaMath-20k
Viewer • Updated • 20.5k • 65 • 1
jasonrqh/Countdown-CoT-20k
Viewer • Updated • 20.5k • 80
jasonrqh/Math-NoCoT-20k
Viewer • Updated • 20.5k • 58
jasonrqh/Math-CoT-20k
Viewer • Updated • 20.5k • 151 • 6