AI Paper Atlas

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

Random Selection Reveals Implicit Knowledge Consensus in Code Generation

Ren-Biao Liu, Xin-Ye Li, Hui Sun, Yali Du, Jiang-Tian Xue, Ming Li

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning for reasoning data

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 random per-problem solution sampling instead of diversity or difficulty selectorsSFT on LeetCode multi-solution pool · LeetCode test Pass@1 (Qwen2.5-7B)Pass@1: 3.90 -> 6.32 (vs K-Center)Improved
Use K-Center, AST Coverage or IFD ranking instead of random selectionSFT code data selection · LeetCode, Qwen2.5-7BPass@1: 6.32 -> 3.90 / 3.73 / 3.42Hurt
Use K-Means or Kernel Herding instead of randomSFT code data selection · LeetCode, Qwen2.5-Coder-7B and DeepSeek-Coder-6.7BPass@1: 5.57 -> 6.14 (K-Means, Coder); 3.68 -> 4.56 (Herding, DeepSeek)Mixed
Use random 11-solution subset instead of the full datasetSFT · LeetCode test Pass@1 (Llama-3.1-8B)Pass@1: 3.68 -> 4.04Improved
Use complex selectors (Facility Location, K-Means, Kernel Herding) at 32B scaleQwen2.5-32B · LeetCode test Pass@10Pass@10: 31.58 -> 35.53 (Facility Location)Improved
Use K-Means / Facility Location instead of Random on constrained Medium/Hard problemsQwen2.5-7B · 59 tagged Medium/Hard problemsPass@10: 6.78 -> 8.47Improved
Use Kernel Herding instead of Random on Hard problemsQwen2.5-7B · LeetCode Hard (79 tasks)Pass@10: 5.06 -> 6.33Improved
Use cosine embedding distance instead of Levenshtein/Jaccard/syntax distance in K-Center (ablation)K-Center · Train Pass@1, Qwen2.5-7BTrain Pass@1: 49.41 -> 58.69Improved
Increase per-problem selection budget k (ablation)Random selection · Qwen2.5-7BTrain Pass@1 / Test Pass@10Improved
Vary learning rate from 1e-6 to 5e-5 (ablation)Random selection · Qwen2.5-7BTest Pass@1: 3.42 (1e-6) -> 6.32 (1e-5) -> 5.35 (5e-5)Mixed

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