AI Paper Atlas

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

Differential Smoothing Mitigates Sharpening and Improves LLM Reasoning

Jingchu Gai, Guanning Zeng, Huaqing Zhang, Aditi Raghunathan

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
add differential smoothing (DS-GRPO) to GRPOGRPO · Pass@K vs base modelPass@8192 change vs base: -3.80 -> +1.60Improved
add differential smoothing (DS-GRPO) to GRPOGRPO · Pass@1 vs base modelPass@1 change vs base: +9.49 -> +12.73Improved
add differential smoothing to GRPOGRPO · CountdownPass@K: approx 4% for all KImproved
add differential smoothing to GRPOGRPO · Math reasoning (3 models, 5 benchmarks)Pass@1: 0.2%--2.9%Improved
add differential smoothing to GRPOGRPO · Math reasoning (3 models, 5 benchmarks)Pass@64: 0.5%--6.7%Improved
add differential smoothing to GRPOGRPO · Math reasoninginference samples to match Pass@64: 4x inference speedup (k=16)Improved
add DS-GRPO under varied sampling temperature (ablation)GRPO · Countdown (temperature sweep)Pass@K: approx 4%Improved
add DS-GRPO under varied KL coefficient (ablation)GRPO · Countdown (KL coefficient sweep)Pass@K: approx 3.2%Improved
modify both correct and incorrect trajectories rather than only correct (add negative modification) (ablation)DS-GRPO-Positive · Countdown/MathPass@KImproved
modify both correct and incorrect trajectories rather than only incorrect (add positive modification) (ablation)DS-GRPO-Negative · Countdown/MathPass@KImproved
use DS-GRPO instead of GR-PKPO (Pass@K reward)GRPO · CountdownPass@KImproved
use DS-GRPO instead of CISPOGRPO · math reasoning datasetsaccuracyImproved

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