Unverified50% confidenceFactExact time
Sparse rewards combined with long time spans cause gradient-based optimization signals to decay significantly during backpropagation.
1
Sources
50%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedGradient descent iteratively adjusts weights in the opposite direction of the loss function's gradient, causing the prediction error to converge progressively.78% similarVerifiedVanishing gradient refers to the exponential decay of gradient values during backpropagation in deep neural networks, causing parameters near the input layer to receive almost no updates.77% similarUnverifiedReverse mode automatic differentiation can obtain gradients of the loss with respect to all parameters in a single backward pass, with complexity only a constant multiple of the forward computation (typically 2–4×).73% similarUnverifiedThe actual speedup gained from MTP depends on the prediction acceptance rate; if predictions are frequently incorrect and fall back to token-by-token decoding, the acceleration benefit is significantly reduced.72% similarUnverifiedThe positive correlation between base task performance and generalization ability weakens under significant distribution shift.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975637API
curl https://kongchang.com/api/v1/knowledge/claims/975637MCP
get_claim(id=975637)