Trained using https://github.com/tloen/alpaca-lora with removing the lines
old_state_dict = model.state_dict
model.state_dict = (
lambda self, *_, **__: get_peft_model_state_dict(
self, old_state_dict()
)
).__get__(model, type(model))
causing problem.
base_model: yahma/llama-7b-hf data_path: prognosis/medical_qa_alpaca output_dir: ./lora-alpaca batch_size: 128 micro_batch_size: 8 num_epochs: 5 learning_rate: 0.0003 cutoff_len: 512 val_set_size: 0.1 lora_r: 16 lora_alpha: 16 lora_dropout: 0.05 lora_target_modules: ['q_proj', 'k_proj', 'v_proj', 'o_proj'] train_on_inputs: True add_eos_token: False group_by_length: True wandb_project: medical_alpaca_hf wandb_run_name: run_3 wandb_watch: wandb_log_model: resume_from_checkpoint: False prompt template: alpaca
Command used
Finetuning
python finetune.py --base_model 'yahma/llama-7b-hf' --data_path 'prognosis/medical_qa_alpaca' --output_dir './lora-alpaca' --wandb_project 'medical_alpaca_hf' --wandb_run_name 'run_3' --lora_target_modules '[q_proj,k_proj,v_proj,o_proj]' --num_epochs 5 --cutoff_len 512 --group_by_length --val_set_size 0.1 --lora_r=16 --micro_batch_size=8
Generating
python generate.py \
--load_8bit \
--base_model 'yahma/llama-7b-hf' \
--lora_weights 'eswardivi/medical_qa_alpaca' \
--share_gradio
git lfs
curl -s https://packagecloud.io/install/repositories/github/git-lfs/script.deb.sh | sudo bash
sudo apt-get install git-lfs