Unverified50% confidenceBenchmarkExact time
All tested models performed far worse on TranSGrid than on the conventional held-out test set.
1
Sources
50%
Confidence
Long-term
Relevance
9/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedFor cell types or species underrepresented in the training data, the reliability of predictions from such models decreases significantly.67% similarUnverifiedThe user pointed out that these models achieve high scores on benchmarks yet frequently fail at the simplest tasks.63% similarUnverified多重比较问题指同时测试多个模型变体时，即使每个单独检验显著性水平设为0.05，整体假阳性率也会急剧膨胀61% similarUnverified即使只有5-10%的训练样本存在错位，也会导致VLM模型在下游任务上出现显著的性能退化60% similarUnverifiedAn early iteration scored five percentage points lower than the Huihui base model on the IFEval instruction-following benchmark.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/931094API
curl https://kongchang.com/api/v1/knowledge/claims/931094MCP
get_claim(id=931094)