victormustar· @victormustar · X·· 11 小时前AI 评分
Hugging Face Hub 模型张量查看提速20倍
我们让 Hugging Face Hub 上查看模型张量的速度提升了最多 20 倍 ⚡ DeepSeek-V3.2: 9.2s → 0.45s GLM-5.3: 4.0s → 0.57s Qwen2.5-72B: 1.2s → 0.27s
We made inspecting models tensors on the Hugging Face Hub up to 20× faster ⚡
DeepSeek-V3.2: 9.2s → 0.45s
GLM-5.3: 4.0s → 0.57s
Qwen2.5-72B: 1.2s → 0.27s
来源：victormustar · x.com