Verified65% confidenceFactExact time
The learning rate determines how large each step is: too large causes overshooting the valley bottom and divergence, while too small makes training extremely time-consuming.
3
Sources
65%
Confidence
Long-term
Relevance
7/14/2026
First Seen
Sources
不靠数学入门AI：7个认知转变让学习更高效
Related Claims
VerifiedDuring cross-domain fine-tuning, an excessively large learning rate will destroy pre-trained weights, while an excessively small one will result in almost no learning, requiring a careful trade-off.83% similarUnverifiedIn the sampled-gradient setting, when the learning rate η decreases, the expected gain under fixed β vanishes faster than the fluctuation itself, causing the benefit of momentum to be swallowed by noise.80% similarUnverifiedWhen people have too little they work hard to learn more, and when they have too much they become distracted and learn less.79% similarUnverifiedUnder a local quadratic model, a larger batch means fewer but longer update steps, with each step incurring a greater loss on curvature, resulting in a critical batch size that depends on the learning rate, beyond which sample efficiency declines.78% similarUnverifiedThe scale of 600 episodes may lead to overly short individual episodes and knowledge points being split too finely, giving learners a fragmented feeling of "having learned a lot but being unable to string it together."76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/507887API
curl https://kongchang.com/api/v1/knowledge/claims/507887MCP
get_claim(id=507887)