Unverified50% confidenceFactExact time
The reinforcement learning modifications tested in the study did not yield consistent gains, and directly training models to learn to abstain proved unstable.
1
Sources
50%
Confidence
Long-term
Relevance
10/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe study concluded that, under its experimental setup, learned contextual planning is only a weak correlation signal rather than a substitute for strong retrieval.77% similarUnverifiedInverse Reinforcement Learning faces an ill-posedness problem—the same set of behavioral data may correspond to multiple entirely different reward functions77% similarUnverifiedThe improvement ceiling of runtime improvement is constrained by the base model's inherent capabilities, and the model cannot break through the knowledge and reasoning boundaries obtained during pre-training through runtime optimization.77% similarUnverifiedThe teacher model outputs in distillation training rarely include self-correction examples, causing student models to form an implicit belief that "strong models don't make mistakes," leading them to repeat rather than correct when verification fails.76% similarUnverified阿里团队发现强化学习训练不会导致模型其他能力下降，反而会带来全面协同提升，与传统微调中的灾难性遗忘问题形成对比75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1002061API
curl https://kongchang.com/api/v1/knowledge/claims/1002061MCP
get_claim(id=1002061)