Unverified50% confidenceFactExact time
Prior work has pointed out that the modality gap is related to the symmetry of the contrastive loss, the temperature coefficient, and initialization, and is difficult to eliminate by continued training.
1
Sources
50%
Confidence
Long-term
Relevance
10/26/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedExperimental conclusion two: some self-distillation methods give attractive scores at the end of training, but during training they exhibit severe performance fluctuations or even collapse followed by recovery, and these problems cannot be found by looking only at the final metric.73% similarUnverifiedThe understanding of text-to-image models is essentially large-scale statistical association, and generation quality drops significantly when dealing with long-tail concepts that appear rarely in the training set.73% similarUnverifiedGradient descent naturally tends to select flatter, simpler solutions from among the many solutions that perfectly fit the training set, and this implicit regularization effect is closely related to generalization ability.71% similarVerifiedNaïve post-training quantization causes individual weight errors to accumulate and amplify across multiple layers—a phenomenon known as compounding error—and at 2-bit or 1-bit precision the model structure collapses70% similarUnverifiedThe performance gap on TranSGrid appears within the training length range, indicating that productivity alone is insufficient to evaluate systematic generalization.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1098275API
curl https://kongchang.com/api/v1/knowledge/claims/1098275MCP
get_claim(id=1098275)