Unverified50% confidenceFactExact time
When the rank r is between 4 and 64, the number of trainable parameters in LoRA is usually on the order of one-thousandth of the original parameters.
1
Sources
50%
Confidence
Long-term
Relevance
10/31/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedLoRA使可训练参数量通常仅为全参微调的0.1%-1%,可在单张RTX 3090/4090(显存24GB)上微调7B乃至13B级别的模型78% similarUnverified一个rank=4的LoRA文件对于SD1.5模型约只有3-6MB,LoRA设计使训练参数量减少95%以上76% similarUnverifiedLoRA训练中常用的学习率为1e-4到5e-4之间,常用的Network Rank值为8、16、32、64、12870% similarUnverified以 GPT-3 为例,r=4 时 LoRA 可将可训练参数从 175B 降至约 4.7M,减少约 37,000 倍69% similarUnverifiedA 360M-parameter code model trained by the trifero.ai lab under Strife Technologies initially reported 26.8% pass@1 on HumanEval, and after removing instruction data overlapping with the benchmark, the true number was 22.6%.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1119226API
curl https://kongchang.com/api/v1/knowledge/claims/1119226MCP
get_claim(id=1119226)