Unverified50% confidenceFactExact time
GPT-2（355M）在混合精度（FP16）下微调显存占用约2-4GB，单卡T4即可完成全流程
1
Sources
50%
Confidence
Long-term
Relevance
7/15/2026
First Seen
Sources
Related Claims
UnverifiedGPT-OSS原生支持FP4混合精度，无需额外量化即可大幅压缩显存占用82% similarVerifiedGPT-2 comes in four sizes: 117M, 345M, 762M, and 1.5B parameters.80% similarVerifiedGPT-3.5 has a vector dimension of 12,288.75% similarUnverifiedGPT-4o 在保持 128K 上下文能力的同时实现了更快的推理速度和更低的成本75% similarUnverifiedGPT-2拥有15亿参数，每次梯度同步需传输约6GB数据74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/513755API
curl https://kongchang.com/api/v1/knowledge/claims/513755MCP
get_claim(id=513755)