run_ap_mist7.sh

python3 -m accelerate.commands.launch --config_file src/config/config.yaml src/train.py --from_pretrained "mistralai/Mistral-7B-v0.1" --model_config "mistral7b" --dir_data_types "computed" --training_data_dirs "/nfs/dgx/raid/chem/Test_Jsons_16_shuf_v0_43/" --valid_data_dir /nfs/dgx/home/ysu/chem/data/valid --train_batch_size 1 --eval_steps 1218 --save_steps 1218 --gradient_accumulation_steps 8 --max_steps 4872 --dataloader_num_workers 16 --experiment_name "mist7b_test" --checkpoints_root_dir /mnt/sxtn/phil/checkpoints/ --track --track_dir /mnt/sxtn/phil/aim/ --flash_attn --gradient_checkpointing