Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
chg0901 created a comment on an issue on ultralytics/ultralytics
> @Dimasilham7 , i did pip install --upgrade ray and still it didn't work in Kaggle, kindly share how you solved yours I also get this problems with torch2.0.0, ultralytics 8.3.58 and ray-2.40.0
chg0901 created a comment on an issue on pytorch/torchtune
Does this setting or configuration work for other models such as llama series? Mircea Mironenco ***@***.***> 于 2025年1月7日周二 16:55写道： > It likely works on 0.5B because the head_dim is smaller....
chg0901 pushed 1 commit to main chg0901/hands_on_torchtune
- Update README.md with problems and links 6a27355
chg0901 created a repository: chg0901/hands_on_torchtune
https://github.com/chg0901/hands_on_torchtune - torchtune实践
chg0901 opened an issue on pytorch/torchtune
Hugging Face from_pretrained() using merged weights KeyError: 'base_model_name_or_path'
test codes in https://pytorch.org/torchtune/stable/tutorials/e2e_flow.html#use-with-hugging-face-from-pretrained ``` from transformers import AutoModelForCausalLM, AutoTokenizer import trans...
chg0901 created a comment on an issue on FunAudioLLM/CosyVoice
> 插一句，https://github.com/Plachtaa/VITS-fast-fine-tuning 这个虽然功能比较少，速度很快。 用自己训练好的模型，20个汉字左右，第一次进行推理大约要6s(估计主要消耗的时间是模型加载)。 换别的文本(同样字数)再次进行推理，1s(或者更少)。之后一直保持1s左右。 > > 不足之处就是，无法支持中英混合的情况，碰到英文它只能一个一个字...