Thanks for taking time.
I tried fine-tune with my own dataset, following default instruction. However, when I ran inference with the new model, I saw this error:
2024-09-23 22:29:19.347 | INFO | __main__:main:662 - Loading model ... Traceback (most recent call last): model: Union[NaiveTransformer, DualARTransformer] = BaseTransformer.from_pretrained( File "/home/colafly/Repo/fish-speech/fish_speech/models/text2semantic/llama.py", line 325, in from_pretrained config = BaseModelArgs.from_pretrained(str(path)) File "/home/colafly/Repo/fish-speech/fish_speech/models/text2semantic/llama.py", line 77, in from_pretrained data = json.load(f) File "/home/colafly/.pyenv/versions/3.10.14/lib/python3.10/json/__init__.py", line 293, in load return loads(fp.read(), File "/home/colafly/.pyenv/versions/3.10.14/lib/python3.10/codecs.py", line 322, in decode (result, consumed) = self._buffer_decode(data, self.errors, final) UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 64: invalid start byte
any idea what it might be the issue?
I did not modify any of the steps, except reduce batch size to 4 to fit into my GPU. Any help is much appreciated.
python tools/llama/generate.py --text "深呼吸" --checkpoint-path "checkpoints/fish-speech-1.4-yth-lora/model.pth" --num-samples 2 --compile
Thanks for taking time.
I tried fine-tune with my own dataset, following default instruction. However, when I ran inference with the new model, I saw this error:
2024-09-23 22:29:19.347 | INFO | __main__:main:662 - Loading model ... Traceback (most recent call last): model: Union[NaiveTransformer, DualARTransformer] = BaseTransformer.from_pretrained( File "/home/colafly/Repo/fish-speech/fish_speech/models/text2semantic/llama.py", line 325, in from_pretrained config = BaseModelArgs.from_pretrained(str(path)) File "/home/colafly/Repo/fish-speech/fish_speech/models/text2semantic/llama.py", line 77, in from_pretrained data = json.load(f) File "/home/colafly/.pyenv/versions/3.10.14/lib/python3.10/json/__init__.py", line 293, in load return loads(fp.read(), File "/home/colafly/.pyenv/versions/3.10.14/lib/python3.10/codecs.py", line 322, in decode (result, consumed) = self._buffer_decode(data, self.errors, final) UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 64: invalid start byteany idea what it might be the issue?
I did not modify any of the steps, except reduce batch size to 4 to fit into my GPU. Any help is much appreciated.
python tools/llama/generate.py --text "深呼吸" --checkpoint-path "checkpoints/fish-speech-1.4-yth-lora/model.pth" --num-samples 2 --compile