The Self-Improvement Paradox: Can Language Models Bootstrap Reasoning Capabilities without External Scaffolding? (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to self-improvement rely on external supervision signals in the form of seed data and/or assistance from third-party models. |
| Approach: | They propose a framework for generating high-quality synthetic question-answer data in a fully autonomous manner. |
| Outcome: | The proposed framework generates high-quality synthetic question-answer data in a fully autonomous manner. |
Similar Papers
Large Language Models Can Self-Improve (2023.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have excellent performance in various tasks, but fine-tuning requires extensive supervision. |
| Approach: | They propose to use a pre-trained Large Language Model to generate rationale-augmented answers for unlabeled questions and fine-tune the LLM using those self-generated solutions as target outputs. |
| Outcome: | The proposed approach improves the general reasoning ability of a 540B-parameter LLM without any ground truth label. |
Can Large Language Models Invent Algorithms to Improve Themselves? (2025.naacl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have shown remarkable performance improvements, but the methods for improving LLMs are still designed by humans. |
| Approach: | They propose a framework which enables LLMs to generate and learn model-improvement algorithms by the seed model. |
| Outcome: | The proposed framework outperforms human-designed methods in model-improving tasks and improves the seed model by 6% and outperformed human-design methods by 4.3% on GSM8k. |
Mitigating Tail Narrowing in LLM Self-Improvement via Socratic-Guided Sampling (2025.naacl-long)
Copied to clipboard
Yiwen Ding, Zhiheng Xi, Wei He, Lizhuoyuan Lizhuoyuan, Yitao Zhai, Shi Xiaowei, Xunliang Cai, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Large language models (LLMs) generate solutions themselves and iteratively train on filtered, high-quality rationales, but performance reaches a ceiling after a few iterations. |
| Approach: | They propose a strategy to improve the efficiency of sampling heavy-tailed data by using Socratic-style guidance signals to help LLMs reasoning with complex queries. |
| Outcome: | The proposed approach is effective on difficult queries and on held-out tasks, while requiring human supervision. |
Vision-Language Models Can Self-Improve Reasoning via Reflection (2025.naacl-long)
Copied to clipboard
| Challenge: | Chain-of-thought (CoT) has been shown to improve the reasoning capability of large language models (LLMs). |
| Approach: | They propose a framework which iteratively enhances the model’s Vision-language Reasoning by Reflecting on CoT Rationales. |
| Outcome: | The proposed framework improves multimodal reasoning on vision-language tasks by 23% to 60% over baselines. |
Self-Explore: Enhancing Mathematical Reasoning in Language Models with Fine-grained Rewards (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Recent studies have shown that large language models can solve complex reasoning tasks with Chain-of-Thought Prompting. |
| Approach: | They propose a training method where the LLM is tasked to explore the first wrong step within the rationale and use such signals as fine-grained rewards for further improvement. |
| Outcome: | The proposed model improves on the GSM8K and MATH test sets by 11.57% and 2.89% on average compared to supervised fine-tuning (SFT). |
Demystifying Mixed Outcomes of Self-Training: Pre-training Analyses on Non-Toy LLMs (2026.findings-eacl)
Copied to clipboard
| Challenge: | Recent studies on self-training report seemingly contradictory outcomes. |
| Approach: | They use OLMo-2 models as non-toy LLMs and perform multiple rounds of continual pre-training using self-generated text with different prompting strategies and data filtering. |
| Outcome: | The proposed model collapse is inherent to the training procedure itself, while self-improvement is likely owes its success to human-designed, strategic synthetic pipelines that inject external intelligence. |
Small Language Models Need Strong Verifiers to Self-Correct Reasoning (2024.findings-acl)
Copied to clipboard
Yunxiang Zhang, Muhammad Khalifa, Lajanugen Logeswaran, Jaekyeom Kim, Moontae Lee, Honglak Lee, Lu Wang
| Challenge: | Existing studies show that large language models can self-correct their outputs by generating a critique and revising it based on the critique. |
| Approach: | They propose a pipeline that prompts small language models to collect self-correction data that supports the training of self-refinement abilities. |
| Outcome: | The proposed pipeline improves the self-correction abilities of two models on five datasets spanning math and commonsense reasoning. |
Meta-Rewarding Language Models: Self-Improving Alignment with LLM-as-a-Meta-Judge (2025.emnlp-main)
Copied to clipboard
Tianhao Wu, Weizhe Yuan, Olga Golovneva, Jing Xu, Yuandong Tian, Jiantao Jiao, Jason E Weston, Sainbayar Sukhbaatar
| Challenge: | Existing methods for improving large language models have focused on improving model responses rather than judgment capabilities, resulting in rapid saturation during iterative training. |
| Approach: | They propose an iterative Meta-Rewarding step where the model judges its own judgements and uses that feedback to refine its judgment skills. |
| Outcome: | The proposed model improves Llama-3-8B-Instruct from 22.9% to 39.4% on AlpacaEval 2 and 20.6% to 29.1% on Arena-Hard. |
S3Prompt: Instructing the Model with Self-calibration, Self-recall and Self-aggregation to Improve In-context Learning (2024.lrec-main)
Copied to clipboard
| Challenge: | Large language models have limitations in practical applications, such as unsupervised generation and recall of in-context examples. |
| Approach: | They propose a self-calibration, self-recall and self-aggregation prompt pipeline to solve these problems. |
| Outcome: | The proposed pipeline improves the performance of large language models without annotating datasets and model parameter updates. |
Advancing Large Language Model Attribution through Self-Improving (2024.emnlp-main)
Copied to clipboard
Lei Huang, Xiaocheng Feng, Weitao Ma, Liang Zhao, Yuchun Fan, Weihong Zhong, Dongliang Xu, Qing Yang, Hongtao Liu, Bing Qin
| Challenge: | Teaching large language models to generate text with citations to evidence sources requires high-quality attribution data, which is costly and labor-intensive. |
| Approach: | They propose a framework for iteratively improving the attribution capability of large language models (LLMs) by attributing output to verifiable sources. |
| Outcome: | Experiments on three open-domain question-answering datasets show that START improves in aggregating information across multiple sources. |