From fc05e59d747ff1cefc6031b08450487a39e653a5 Mon Sep 17 00:00:00 2001 From: roc Date: Fri, 26 Apr 2024 19:54:52 +0800 Subject: [PATCH] update at 2024-04-26 19:54:52 --- content/cases/llama3.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/content/cases/llama3.md b/content/cases/llama3.md index 161bcbf..52dcd5a 100644 --- a/content/cases/llama3.md +++ b/content/cases/llama3.md @@ -58,6 +58,10 @@ pulling manifest Error: pull model manifest: Get "https://registry.ollama.ai/v2/library/llama3/manifests/70b": dial tcp 172.67.182.229:443: i/o timeout ``` +### 70b 的速度非常慢 + +70b 是 700 亿参数的大模型，使用 CPU 运行不太现实，使用 GPU 也得显存足够大，实测用 32G 显存的显卡运行速度也非常慢，建议至少 40G（比如 A100）。 + ## 参考资料 * Llama3 模型库：https://ollama.com/library/llama3