Unverified50% confidenceCautionExact time
Multi-task learning faces the risk of negative transfer, where joint training can harm the performance of individual tasks when the gradient directions between tasks conflict.
1
Sources
50%
Confidence
Long-term
Relevance
9/6/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedSimply stacking the number of RL tasks does not lead to better training outcomes, and unsolvable tasks instead drag down overall training quality.71% similarUnverified分配给预训练的算力越多，下游任务的pass@1奖励就越高，预训练损失越低模型在后续RL中获得的收益越大70% similarUnverified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70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/866093API
curl https://kongchang.com/api/v1/knowledge/claims/866093MCP
get_claim(id=866093)