[KongchangAI]
Partially Verified80% confidenceBenchmarkExact time

以 65B 参数的 LLaMA 模型为例,传统全量微调需要 780GB 以上显存,而 QLoRA 仅需约 48GB

4
Sources
80%
Confidence
Long-term
Relevance
7/7/2026
First Seen

Sources

Related Claims

Cite This Claim

Stable URI
https://kongchang.com/claim/172555
API
curl https://kongchang.com/api/v1/knowledge/claims/172555
MCP
get_claim(id=172555)