Unverified50% confidencePredictionExact time
If LLM judges that are blind to omissions are widely used for scoring, benchmarking, or guiding model training (e.g., RLHF, RLAIF), the entire optimization loop may be misled, causing models to learn to generate content that appears complete but is actually incomplete.
1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen
Sources
Related Entities
Related Claims
Unverified模型在未收到明确指令时输出图片，意味着系统提示词设计或RLHF训练中存在激励偏差79% similarUnverifiedIn scenarios with insufficient training data coverage, LLMs generate outputs that appear plausible but may be incorrect based on statistical patterns, rather than acknowledging uncertainty.78% similarUnverifiedThe systematic process for debugging a non-converging model is to first verify on a very small data subset whether the model can overfit; if it cannot overfit even on small data, it indicates a bug in the model or training code.78% similarUnverifiedRLHF training causes models to develop 'sycophancy,' meaning the model would rather fabricate information than admit ignorance.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/909756API
curl https://kongchang.com/api/v1/knowledge/claims/909756MCP
get_claim(id=909756)