Unverified50% confidenceFactExact time
Weight compression during the post-training stage redistributes the error burden across different demographic groups by directly modifying model parameters, thereby affecting fairness.
1
Sources
50%
Confidence
Long-term
Relevance
9/25/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedTraditional draft models are trained on narrow distributions targeting a single model, causing acceptance rates to drop sharply when workloads change, and requiring the entire pre-training process to be redone each time a target model is replaced.75% similarUnverifiedDuring the post-training phase, an excessively low loss may correspond to overfitting, where the model memorizes the training data and loses generalization ability.73% similarUnverifiedIn semantic segmentation tasks, deep supervision is typically implemented as a weighted sum of the main loss and auxiliary losses, with the auxiliary loss weights often gradually decayed in the later stages of training.73% similarUnverifiedUnconstrained base prompts cause the model to fall back to the average-state output most common in its training data, while deeply personalized prompts can significantly improve output uniqueness.72% similarVerifiedMoE training typically requires the introduction of auxiliary loss to enforce load balancing; otherwise, a Matthew effect forms where a minority of Experts become increasingly dominant, causing most Experts to degrade into dead code.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/949153API
curl https://kongchang.com/api/v1/knowledge/claims/949153MCP
get_claim(id=949153)