stanford_alpaca
stanford_alpaca copied to clipboard
tatsu-lab
Code and documentation to train Stanford's Alpaca models, and generate the data.
Thanks for your good job，and does the model support the task in Chinese?
Can you please send how to train on v100? i have tried multiple attempts always lead to OOM Error. Also Tried with batch size 1
We implement a simple adapter codebase for finetuning llama in the following github link: https://github.com/ZrrSkywalker/LLaMA-Adapter We are going to support larger models and try multimodal input.
Hi, thank you for the nice repo! while looking at the train.py I recognized that `[PAD]` is added as a special token to the tokenizer and the model embeddings. When...
Not sure whether it affected the training data but there is a spelling mistake on prompt.txt at line 5. It should be "instructions" instead of "instrucitons".
Hi all friends, welcome to join in QQ chat group and discuss all problems and experience. The QQ chat group number is: 397447632
The website (https://alpaca-ai.ngrok.io/) gives an error like below. 
I have completed fine-tuning in 8 A100 gpus, when loading the ft model by "model=model.to("cuda")" it appeared OOM. And I have set 'os.environ['CUDA_VISIBLE_DEVICES'] = "0,1,2,3,4,5,6,7"' during generation.
I found a code error, where the label assignment doesn't work, please ref：https://github.com/tatsu-lab/stanford_alpaca/blob/eb5b171d9b103a12a8e14e0edca9cbc45fe1d512/train.py#L132
