You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
[FOR CHINESE USERS] 请务必使用英文提交 Issue,否则会被关闭。谢谢!:)
Please do not modify this template :) and fill in all the required fields.
Cloud or Self Hosted
Self Hosted (Source)
Steps to reproduce
I click start.bat and go to inference configuration.
for Decoder Model Path path i select:
checkpoints\fish-speech-1.4\firefly-gan-vq-fsq-8x1024-21hz-generator.pth
Decoder Model Config:
firefly_gan_vq
LLAMA Model Path:
checkpoints\fish-speech-1.4
Compile model:
No
click Open Inference Server
I get this
"HF_ENDPOINT: https://huggingface.co"
"NO_PROXY: "
model.pth 已存在,跳过下载。
README.md 已存在,跳过下载。
special_tokens_map.json 已存在,跳过下载。
tokenizer_config.json 已存在,跳过下载。
tokenizer.json 已存在,跳过下载。
config.json 已存在,跳过下载。
firefly-gan-vq-fsq-4x1024-42hz-generator.pth 已存在,跳过下载。
ffmpeg.exe 已存在,跳过下载。
ffprobe.exe 已存在,跳过下载。
asr-label-win-x64.exe 已存在,跳过下载。
Debug: flags =
Next launch the page...
['', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1\\fish_speech\\webui', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1\\fishenv\\env\\python310.zip', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1\\fishenv\\env\\DLLs', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1\\fishenv\\env\\lib', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1\\fishenv\\env', 'C:\\Users\\J__mo\\Downloads\\fish-speech-1.2.1\\fish-speech-1.2.1\\fishenv\\env\\lib\\site-packages', '__editable__.fish_speech-0.1.0.finder.__path_hook__']
You are in C:\Users\J__mo\Downloads\fish-speech-1.2.1\fish-speech-1.2.1
Running on local URL: http://127.0.0.1:7860
To create a public link, set `share=True` in `launch()`.
2024-09-11 20:50:33.852 | INFO | __main__:clean_infer_cache:146 - C:\Users\J__mo\AppData\Local\Temp\gradio was not found
2024-09-11 20:50:49.634 | INFO | __main__:<module>:584 - Loading Llama model...
2024-09-11 20:50:55.998 | INFO | tools.llama.generate:load_model:352 - Restored model from checkpoint
2024-09-11 20:50:55.999 | INFO | tools.llama.generate:load_model:356 - Using DualARTransformer
2024-09-11 20:50:56.001 | INFO | __main__:<module>:591 - Llama model loaded, loading VQ-GAN model...
Traceback (most recent call last):
File "C:\Users\J__mo\Downloads\fish-speech-1.2.1\fish-speech-1.2.1\tools\webui.py", line 593, in <module>
decoder_model = load_decoder_model(
File "C:\Users\J__mo\Downloads\fish-speech-1.2.1\fish-speech-1.2.1\tools\vqgan\inference.py", line 41, in load_model
result = model.load_state_dict(state_dict, strict=False)
File "C:\Users\J__mo\Downloads\fish-speech-1.2.1\fish-speech-1.2.1\fishenv\env\lib\site-packages\torch\nn\modules\module.py", line 2191, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for FireflyArchitecture:
size mismatch for quantizer.residual_fsq.rvqs.0.project_in.weight: copying a param with shape torch.Size([4, 64]) from checkpoint, the shape in current model is torch.Size([4, 128]).
size mismatch for quantizer.residual_fsq.rvqs.0.project_out.weight: copying a param with shape torch.Size([64, 4]) from checkpoint, the shape in current model is torch.Size([128, 4]).
size mismatch for quantizer.residual_fsq.rvqs.0.project_out.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for quantizer.residual_fsq.rvqs.1.project_in.weight: copying a param with shape torch.Size([4, 64]) from checkpoint, the shape in current model is torch.Size([4, 128]).
size mismatch for quantizer.residual_fsq.rvqs.1.project_out.weight: copying a param with shape torch.Size([64, 4]) from checkpoint, the shape in current model is torch.Size([128, 4]).
size mismatch for quantizer.residual_fsq.rvqs.1.project_out.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for quantizer.residual_fsq.rvqs.2.project_in.weight: copying a param with shape torch.Size([4, 64]) from checkpoint, the shape in current model is torch.Size([4, 128]).
size mismatch for quantizer.residual_fsq.rvqs.2.project_out.weight: copying a param with shape torch.Size([64, 4]) from checkpoint, the shape in current model is torch.Size([128, 4]).
size mismatch for quantizer.residual_fsq.rvqs.2.project_out.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for quantizer.residual_fsq.rvqs.3.project_in.weight: copying a param with shape torch.Size([4, 64]) from checkpoint, the shape in current model is torch.Size([4, 128]).
size mismatch for quantizer.residual_fsq.rvqs.3.project_out.weight: copying a param with shape torch.Size([64, 4]) from checkpoint, the shape in current model is torch.Size([128, 4]).
size mismatch for quantizer.residual_fsq.rvqs.3.project_out.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
The 1.2 model loads and works fine just not 1.4
✔️ Expected Behavior
For the inference server to launch with the new model.
❌ Actual Behavior
It crashed.
The text was updated successfully, but these errors were encountered:
Try the command with your virtual environment activated. python tools/webui.py
please run the command under the fish-speech folder.
Also, please ensure your code version, you can't run the V1.4 model with V1.2 source code.
Self Checks
Cloud or Self Hosted
Self Hosted (Source)
Steps to reproduce
I click start.bat and go to inference configuration.
for Decoder Model Path path i select:
checkpoints\fish-speech-1.4\firefly-gan-vq-fsq-8x1024-21hz-generator.pth
Decoder Model Config:
firefly_gan_vq
LLAMA Model Path:
checkpoints\fish-speech-1.4
Compile model:
No
click Open Inference Server
I get this
The 1.2 model loads and works fine just not 1.4
✔️ Expected Behavior
For the inference server to launch with the new model.
❌ Actual Behavior
It crashed.
The text was updated successfully, but these errors were encountered: