QLoRAで学習したつくよみちゃん LoRAモデル

学習パラメータ

!python qlora.py \
    --model_name cyberagent/calm2-7b-chat \
    --output_dir "./output/calm2-7b-chat-tsukuyomi-chan" \
    --dataset "alpaca" \
    --max_steps 1500 \
    --use_auth \
    --logging_steps 100 \
    --save_strategy steps \
    --data_seed 42 \
    --save_steps 300 \
    --save_total_limit 100 \
    --max_new_tokens 5000 \
    --dataloader_num_workers 1 \
    --group_by_length \
    --logging_strategy steps \
    --remove_unused_columns False \
    --do_train \
    --lora_r 64 \
    --lora_alpha 16 \
    --lora_modules all \
    --double_quant \
    --quant_type nf4 \
    --bf16 \
    --bits 4 \
    --warmup_ratio 0.03 \
    --lr_scheduler_type constant \
    --gradient_checkpointing \
    --source_max_len 1024 \
    --target_max_len 512 \
    --per_device_train_batch_size 4 \
    --gradient_accumulation_steps 4 \
    --eval_steps 187 \
    --learning_rate 0.0001 \
    --adam_beta2 0.999 \
    --max_grad_norm 0.3 \
    --lora_dropout 0.1 \
    --weight_decay 0.0 \
    --seed 0 \
    --load_in_4bit \
    --use_peft \
    --batch_size 4 \
    --gradient_accumulation_steps 2
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API: The model has no library tag.