AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

How to Fine-Tune a Reasoning Model? A Teacher–Student Cooperation Framework to Synthesize Student-Consistent SFT Data

Zixian Huang, Kaichen Yang, Xu Huang, Feiyang Hao, Qiming Ge, Bowen Li, He Du, Kai Chen, Qipeng Guo

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning for reasoning data · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use TESSY teacher-student cooperative synthesis instead of Teacher-Only dataQwen3-8B SFT · OJBenchpass@1: 8.73 -> 25.43Improved
use TESSY instead of Teacher-Only dataQwen3-8B SFT · LiveCodeBench-Propass@1: 22.10 -> 36.69Improved
SFT on Teacher-Only (GPT-OSS-120B) dataQwen3-8B · OJBenchpass@1: 18.75 -> 8.73Hurt
SFT on Self-Distillation dataQwen3-8B · LiveCodeBench-Propass@1: 25.35 -> 19.26Hurt
SFT on Reject-Sampling dataQwen3-8B · LiveCodeBench-Propass@1: 25.35 -> 26.06No clear effect
use TESSY instead of Teacher-Only with DeepSeek-R1-Distill-Llama-8B studentDeepSeek-R1-Distill-Llama-8B · LiveCodeBench-V5pass@1: 38.32 -> 52.10Improved
use student (not teacher) to generate final answer in TESSY (ablation)TESSY · LiveCodeBench-Pro / OJBenchpass@1: 12.33% and 13.58% drops when teacher generates answerImproved
train boundary predictors on 100K instead of 500 samples (ablation)TESSY · OJBench (40K samples)pass@1: 23.98 -> 26.51Improved
add independent system prompt isolation to Teacher-Only training (ablation)Teacher-Only SFT · OJBench / LiveCodeBench-Propass@1: +2.55% and +5.28%Improved
use LoRA (rank 8, alpha 16) instead of full fine-tuning (ablation)Qwen3-8B SFT · LiveCodeBench-Pro / OJBenchpass@1Hurt
start SFT from reasoning model Qwen3-8B instead of Qwen3-8B-Base (ablation)TESSY SFT · OJBench / LiveCodeBench-Propass@1: base lags by 17.67%Improved
use TESSY on Qwen3-30B-A3B studentQwen3-30B-A3B · LiveCodeBench-Pro / OJBenchpass@1: +6.37% and +10.13%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy