PKU-YuanGroup / Open-Sora-Plan

This project aim to reproduce Sora (Open AI T2V model), we wish the open source community contribute to this project.
MIT License
11.25k stars 1k forks source link

opensora/serve/gradio_web_server.py 的问题 #371

Open fallbernana123456 opened 1 month ago

fallbernana123456 commented 1 month ago

在opensora/serve/gradio_web_server.py 里引用了 text_encoder = MT5EncoderModel.from_pretrained("/storage/ongoing/new/Open-Sora-Plan/cache_dir/mt5-xxl", cache_dir=args.cache_dir, low_cpu_mem_usage=True, torch_dtype=weight_dtype) tokenizer = AutoTokenizer.from_pretrained("/storage/ongoing/new/Open-Sora-Plan/cache_dir/mt5-xxl", cache_dir=args.cache_dir)

这个/storage/ongoing/new/Open-Sora-Plan/cache_dir/mt5-xxl 在哪里获取?是指的 google/mt5-xxl 吗?

LinB203 commented 1 month ago

是的。

fallbernana123456 commented 1 month ago

是的。 google/mt5-xxl 有50G。我的gpu只有32G。那1.2的需要gpu大小最小要多少?能用T5-XXL 替代吗?

LinB203 commented 3 weeks ago

是的。 google/mt5-xxl 有50G。我的gpu只有32G。那1.2的需要gpu大小最小要多少?能用T5-XXL 替代吗?

40G,不能,只能用mt5-xxl。