Unverified50% confidenceFactExact time
Chinchilla论文修正了预训练中计算与数据的最优比例关系
1
Sources
50%
Confidence
Long-term
Relevance
8/25/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedDeepMind's Chinchilla research (2022) pointed out that optimal training should keep data volume and parameter count growing at roughly a 1:1 ratio77% similarUnverifiedThe Chinchilla law states that under a fixed compute budget, appropriately reducing model size and increasing training data volume often yields better performance.75% similarUnverifiedThe Chinchilla scaling law states that the optimal number of training tokens should be proportional to the number of model parameters, meaning that when parameters are doubled, training data and computational cost should at least double as well.73% similarVerifiedIn 2022, DeepMind's Chinchilla paper pointed out that under the same compute budget, model size and training data volume should be scaled up proportionally.73% similarUnverified根据Chinchilla定律及其后续研究，模型性能与训练计算量之间存在幂律关系71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/799918API
curl https://kongchang.com/api/v1/knowledge/claims/799918MCP
get_claim(id=799918)