Unverified50% confidenceFactExact time
Perplexity is the standard metric for language modeling, equal to the exponent of the average negative log-likelihood, and lower is better; zero-shot accuracy refers to the accuracy of a model answering directly in the task format without downstream fine-tuning.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedPerplexity is the most commonly used metric for measuring compression loss in language models, reflecting the model's prediction uncertainty on test corpora, with lower values indicating better language understanding.78% similarUnverifiedThere is a gap between the output confidence of language models and their actual accuracy, a phenomenon known in machine learning as "miscalibration."71% similarUnverified量化模型的困惑度可能仅上升0.1-0.5，但在特定下游任务（如代码生成、数学推理）上的表现可能出现显著退化71% similarUnverifiedGiving models vague instructions leads to vastly different outputs with low comparability, so evaluations should use very specific prompts.71% similarUnverifiedPerplexity is the most commonly cited metric for measuring quantization quality; a lower value indicates stronger predictive ability, and it is typically measured using standard datasets such as WikiText-2.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114900API
curl https://kongchang.com/api/v1/knowledge/claims/1114900MCP
get_claim(id=1114900)