Unverified50% confidenceFactExact time
Using CVRP as a controlled proxy, the research team systematically identified three factors behind cross-scale generalization degradation: node density shift, attention diffusion, and dilution of graph context by mean pooling.
1
Sources
50%
Confidence
Long-term
Relevance
10/21/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe researchers propose the attention dilution effect hypothesis to explain why batching reduces overcorrection.69% similarUnverifiedMixed-precision quantization retains higher bit precision for key weights in attention layers while applying more aggressive compression to feed-forward layers with higher redundancy67% similarUnverifiedThe study observed consistent patterns in both controlled and real collective settings: PKS decreases as informative context accumulates.66% similarUnverifiedIn Transformer models, the first and last few layers are more sensitive to quantization, the middle layers have higher redundancy, and attention layers and feed-forward network layers differ in their tolerance to quantization.66% similarUnverifiedThe researcher observed three divergence patterns: differences being absorbed and then converging, differences persistently leading to different outcomes, and divergence—reconvergence—divergence again.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1079403API
curl https://kongchang.com/api/v1/knowledge/claims/1079403MCP
get_claim(id=1079403)