Unverified50% confidenceFactExact time
In the safety report released alongside Llama 2, Meta acknowledged that safety fine-tuning caused the model's scores to decline on certain benchmarks.
1
Sources
50%
Confidence
Long-term
Relevance
9/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedMeta adopted a relatively low-key strategy for the release of LLaMA 3, making it publicly available quickly after confirming that performance met the bar.70% similarUnverified基准测试面临饱和（Saturation）和数据污染（Data Contamination）两个系统性挑战，当多个主流模型得分超过95%时测试集失去区分能力65% similarUnverified在MATH数据集的Level 5竞赛级难题上，前沿模型准确率仍低于60%63% similarUnverifiedMultiple benchmark tests in 2023 (such as the HaHackathon dataset) showed that top LLMs still have significantly lower accuracy than ordinary human annotators on sarcasm recognition tasks62% similarUnverifiedThe average evaluation metrics of a new LLM may be better than those of an older model, but regressions can occur in niche critical scenarios.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/845777API
curl https://kongchang.com/api/v1/knowledge/claims/845777MCP
get_claim(id=845777)