腾讯混元-DiT 文生图

原创已于 2025-05-04 10:55:10 修改 · 527 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#人工智能

于 2025-05-03 16:32:33 首次发布

混元-DiT所需的模型大小一共是41G

https://huggingface.co/Tencent-Hunyuan/HunyuanDiT

https://colab.research.google.com/

HunyuanDiT_jupyter.ipynb

%cd /content
!GIT_LFS_SKIP_SMUDGE=1 git clone -b dev https://github.com/camenduru/HunyuanDiT
%cd /content/HunyuanDiT

!pip install gradio==4.31.0 loguru==0.7.2 diffusers==0.21.2 timm==0.9.5 einops==0.7.0 accelerate==0.29.3

!apt -y install -qq aria2
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/clip_text_encoder/config.json -d /content/HunyuanDiT/ckpts/t2i/clip_text_encoder -o config.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/clip_text_encoder/pytorch_model.bin -d /content/HunyuanDiT/ckpts/t2i/clip_text_encoder -o pytorch_model.bin
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/model/pytorch_model_ema.pt -d /content/HunyuanDiT/ckpts/t2i/model -o pytorch_model_ema.pt
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/model/pytorch_model_module.pt -d /content/HunyuanDiT/ckpts/t2i/model -o pytorch_model_module.pt
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/mt5/config.json -d /content/HunyuanDiT/ckpts/t2i/mt5 -o config.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/mt5/generation_config.json -d /content/HunyuanDiT/ckpts/t2i/mt5 -o generation_config.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/mt5/pytorch_model.bin -d /content/HunyuanDiT/ckpts/t2i/mt5 -o pytorch_model.bin
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/mt5/special_tokens_map.json -d /content/HunyuanDiT/ckpts/t2i/mt5 -o special_tokens_map.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/mt5/spiece.model -d /content/HunyuanDiT/ckpts/t2i/mt5 -o spiece.model
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/mt5/tokenizer_config.json -d /content/HunyuanDiT/ckpts/t2i/mt5 -o tokenizer_config.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/sdxl-vae-fp16-fix/config.json -d /content/HunyuanDiT/ckpts/t2i/sdxl-vae-fp16-fix -o config.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/sdxl-vae-fp16-fix/diffusion_pytorch_model.bin -d /content/HunyuanDiT/ckpts/t2i/sdxl-vae-fp16-fix -o diffusion_pytorch_model.bin
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/resolve/main/t2i/sdxl-vae-fp16-fix/diffusion_pytorch_model.safetensors -d /content/HunyuanDiT/ckpts/t2i/sdxl-vae-fp16-fix -o diffusion_pytorch_model.safetensors
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/tokenizer/special_tokens_map.json -d /content/HunyuanDiT/ckpts/t2i/tokenizer -o special_tokens_map.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/tokenizer/tokenizer_config.json -d /content/HunyuanDiT/ckpts/t2i/tokenizer -o tokenizer_config.json
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/tokenizer/vocab.txt -d /content/HunyuanDiT/ckpts/t2i/tokenizer -o vocab.txt
!aria2c --console-log-level=error -c -x 16 -s 16 -k 1M https://huggingface.co/Tencent-Hunyuan/HunyuanDiT/raw/main/t2i/tokenizer/vocab_org.txt -d /content/HunyuanDiT/ckpts/t2i/tokenizer -o vocab_org.txt

!python app/hydit_app.py --no-enhance --lang en

!pip install huggingface-hub==0.25.2

%cd /content/HunyuanDiT
!pip uninstall deepspeed transformers accelerate -y  
!pip install -r requirements.txt  
!pip install transformers==4.37.2 accelerate==0.28.0 deepspeed==0.14.4 
!python app/hydit_app.py --no-enhance --lang zh

报错一直无法解决

!pip install huggingface-hub==0.25.2
!pip install pydantic==2.10.6
!pip install --upgrade gradio==4.44.1

%cd /content/HunyuanDiT
!python app/hydit_app.py --no-enhance --lang zh