stanford_alpaca
stanford_alpaca copied to clipboard
tatsu-lab
Code and documentation to train Stanford's Alpaca models, and generate the data.
``` Traceback (most recent call last): File "/home/ubuntu/stanford_alpaca/train.py", line 231, in train() File "/home/ubuntu/stanford_alpaca/train.py", line 225, in train trainer.train() File "/home/ubuntu/anaconda3/envs/lama/lib/python3.10/site-packages/transformers-4.27.0.dev0-py3.10.egg/transformers/trainer.py", line 1628, in train return inner_training_loop( File "/home/ubuntu/anaconda3/envs/lama/lib/python3.10/site-packages/transformers-4.27.0.dev0-py3.10.egg/transformers/trainer.py", line...
https://github.com/tatsu-lab/stanford_alpaca/blob/73cac8be49a66ca5d159ee9199428804e1e6aabe/generate_instruction.py#L155
how to use the training data to fine-tune the open source chat-glm
``` F:\!DEV\stanford_alpaca>torchrun --nproc_per_node=4 --master_port=2233 train.py --model_name_or_path "facebook/opt-6.7b" --data_path ./alpaca_data.json --bf16 True --output_dir results --num_train_epochs 3 --per_device_train_batch_size 4 --per_device_eval_batch_size 4 --gradient_accumulation_steps 8 --evaluation_strategy "no" --save_strategy "steps" --save_steps 2000 --save_total_limit 1 --learning_rate...
As mentioned in the repo, Hugging Face's transformers library by installing from a particular fork (i.e. this https://github.com/huggingface/transformers/pull/21955 to be merged). So updating requirements.txt to install that version of transformers...
Uses link references to fix the links to the two cited papers. Keeps the syntax similar to LaTeX, no changes to rendered document except for the links inside "[1]" and...
Hi there, when I get it right Alpaca is a pretrained, locally running AI. Is it possible to pass new input to Alpaca, f.e. let Alpaca summarize a new text?
I just get access to Meta LLaMa model parameters. Whats to do next? Thanks lol
instrucitons -> instructions necssary -> necessary "a instr.." -> "an instr..."
Is there a way to visualize the activation status of each layer in the llama model? I want to observe how the language model responds to different types of problems,...
