Unverified50% confidenceFactExact time
Which factors in the training process drive the rise of α, and how different training objectives, data distributions, and model architectures affect this slope, remain to be answered.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedFrontier models exhibit emergent capabilities, meaning they suddenly display abilities not explicitly included in the training objective once a certain parameter-scale threshold is reached.75% similarUnverifiedDifferent training objectives (such as contrastive learning and masked language modeling) may produce different α ranges, offering a new dimension for comparing different training paradigms.73% similarUnverifiedIn LoRA training, Alpha functions as a scaling factor; the effective learning rate approximates Alpha/rank, and when the two are equal the scaling ratio is 1.72% similarUnverifiedDuring large model training, operations such as gradient accumulation, momentum updates, and weight decay are essentially additions, and when weights are large while gradient updates are extremely small, some learning signals may be silently discarded.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023611API
curl https://kongchang.com/api/v1/knowledge/claims/1023611MCP
get_claim(id=1023611)