Unverified50% confidenceFactExact time
The study uses English GPT-2 for nine generations of self-training, fine-tuning each generation on the previous generation's outputs and varying the sampling temperature of the training data; all models are measured with the same fixed sampler to ensure fair comparison.
1
Sources
50%
Confidence
Long-term
Relevance
11/9/2026
First Seen
Sources
Related Entities
Related Claims
Unverified当比较两个具体训练模型的预测行为差异时应固定种子使用同一测试集运行McNemar检验,当评估方法整体优越性时应用多种子重复实验配合配对t检验或非参数检验74% similarVerifiedWhen evaluating a model, data is usually split into a training set and a test set, with the test set used to measure the true level of generalization.69% similarUnverifiedDuring model selection, it is necessary to confirm that the similarity objective function used during model training is consistent with the metric used during retrieval; otherwise, systematic bias will occur.69% similarUnverifiedThe researchers used a formal grammar lens benchmark to test GPT-5.5 and its Pro-Medium, Pro-High, and Pro-XHigh reasoning-level variants, and found that the model's acceptance rate on the test data increases with reasoning level.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1149958API
curl https://kongchang.com/api/v1/knowledge/claims/1149958MCP
get_claim(id=1149958)