Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

SPACE: Noise Contrastive Estimation Stabilizes Self-Play Fine-Tuning for Large Language Models

About

Self-play fine-tuning has demonstrated promising abilities in adapting large language models (LLMs) to downstream tasks with limited real-world data. The basic principle is to iteratively refine the model with real samples and synthetic ones generated from itself. However, the existing methods primarily focus on the relative gaps between the rewards for two types of data, neglecting their absolute values. Through theoretical analysis, we identify that the gap-based methods suffer from unstable evolution, due to the potentially degenerated objectives. To address this limitation, we introduce a novel self-play fine-tuning method, namely Self-PlAy via Noise Contrastive Estimation (SPACE), which leverages noise contrastive estimation to capture the real-world data distribution. Specifically, SPACE treats synthetic samples as auxiliary components, and discriminates them from the real ones in a binary classification manner. As a result, SPACE independently optimizes the absolute reward values for each type of data, ensuring a consistently meaningful objective and thereby avoiding the instability issue. Theoretically, we show that the optimal solution of the objective in SPACE aligns with the underlying distribution of real-world data, and SPACE guarantees a provably stable convergence to the optimal distribution. Empirically, we show that SPACE significantly improves the performance of LLMs over various tasks, and outperforms supervised fine-tuning that employs much more real-world samples. Compared to gap-based self-play fine-tuning methods, SPACE exhibits remarkable superiority and stable evolution.

Yibo Wang, Qing-Guo Chen, Zhao Xu, Weihua Luo, Kaifu Zhang, Lijun Zhang• 2025

Related benchmarks

TaskDatasetResultRank
Commonsense ReasoningHellaSwag
Accuracy83.64
1460
Language UnderstandingMMLU
Accuracy58.2
756
ReasoningBBH--
507
Instruction FollowingIFEval--
292
Commonsense ReasoningWinoGrande
Accuracy74.11
231
Question AnsweringARC
Accuracy64.85
154
Language UnderstandingMMLU-Pro
Accuracy28.81
70
HallucinationTruthfulQA
Score48.74
42
Aggregate PerformanceMulti-task Evaluation Suite
Average Performance49
21
Large Language Model EvaluationHuggingFace Open LLM Leaderboard
ARC65.96
21
Showing 10 of 12 rows

Other info

Follow for update