Unverified50% confidenceFactExact time
The core mechanism of Self-Refine is to have the large language model critically evaluate its own initial output, then generate an improved version based on the evaluative feedback, iterating until requirements are met or the iteration limit is reached, with the entire process relying solely on a single model's reasoning capability and requiring no external reward model or human feedback.
1
Sources
50%
Confidence
Long-term
Relevance
8/31/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe self-improvement mechanism achieves behavioral optimization primarily through context engineering at the inference stage, which differs from online fine-tuning that involves updating model parameters.74% similarUnverifiedSelf-Refine（Madaan等，2023）让模型生成初始输出后以批评者身份评估并给出改进建议再重新生成73% similarUnverifiedSelf-reflection is achieved by having the model evaluate the quality of its own output and iteratively improve it; the Reflexion framework demonstrates that an agent can learn from failed attempts and adjust its strategy in the next iteration.72% similarUnverifiedLarge model output generation uses an autoregressive approach to produce tokens one by one, requiring recalculation of attention over all preceding context with each token generated.72% similarUnverifiedBy introducing external feedback, ReAct allows models to think while doing and adjust while thinking, effectively mitigating the drift problem of pure reasoning.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/831915API
curl https://kongchang.com/api/v1/knowledge/claims/831915MCP
get_claim(id=831915)