Unverified50% confidenceFactExact time
In distributed training, a small number of nodes whose kernel parameters are inconsistent with the rest can drag down the throughput of the entire training job, creating bottleneck nodes.
1
Sources
50%
Confidence
Long-term
Relevance
9/23/2026
First Seen
Sources
Related Entities
Related Claims
Unverified针对特定工具做专项强化训练的副作用是对非训练分布工具的泛化能力下降，学术上称为分布外泛化失败79% similarVerifiedBehavioral cloning suffers from the distribution shift problem: during training the policy only encounters the state distribution covered by expert trajectories, and during deployment the accumulation of small deviations causes the actual state distribution visited to gradually diverge from the training distribution.78% similarUnverifiedDuring the post-training phase, an excessively low loss may correspond to overfitting, where the model memorizes the training data and loses generalization ability.78% similarUnverifiedIn hyperscale training, a single node or link failure can cause the entire training job to be interrupted or rolled back to the previous checkpoint.77% similarUnverifiedA 2017 study by Keskar et al. showed that large-batch training tends to converge to sharp minima corresponding to worse generalization, while small-batch training tends toward flat minima.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/944939API
curl https://kongchang.com/api/v1/knowledge/claims/944939MCP
get_claim(id=944939)