Unverified50% confidenceOpinionExact time
After fine-tuning, the model's responses are generally shorter, less polished in wording, and overall closer to real human expression.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe same model may produce completely different answers when faced with semantically similar but differently worded questions.73% similarUnverifiedLarge language models have a sycophancy characteristic, tending to organize their output according to the inclination of the prompt rather than reflecting factual truth.70% similarUnverifiedPerplexity measures a language model's ability to predict test text; the lower the value, the less surprised the model is by the text.67% similarUnverifiedThe local small model's accuracy decreases in complex scenarios involving mixed Chinese and English or a high density of professional terminology.67% similarUnverifiedCLAUDE.md is a soft constraint that influences large language model behavior through context injection, and the model may forget or ignore its instructions during long conversations due to context window limitations or attention drift.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/913194API
curl https://kongchang.com/api/v1/knowledge/claims/913194MCP
get_claim(id=913194)