Unverified60% confidenceFactExact time
GPT-3 has 175 billion parameters, and storing these weights alone requires hundreds of gigabytes of space
2
Sources
60%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
零基础入门大模型微调：原理、场景与实战路径
Related Claims
UnverifiedThe KV cache for a 128K context window of a GPT-4-level model can occupy tens of gigabytes of VRAM76% similarVerifiedRunning local large models with more than 7 billion parameters typically requires a GPU with at least 16GB of VRAM.75% similarUnverifiedInferencing a 1T-parameter MoE model requires hundreds of gigabytes of VRAM even with BF16 half-precision storage, far exceeding the capacity limit of consumer-grade GPUs.73% similarUnverifiedH100的内存带宽约3.35TB/s，GPT-4规模模型单次生成一个Token需读取约1TB参数数据，理论上每秒最多生成约3个Token70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/607066API
curl https://kongchang.com/api/v1/knowledge/claims/607066MCP
get_claim(id=607066)