private-model / lora-scripts /train_by_toml.sh
ACCC1380's picture
Upload lora-scripts/train_by_toml.sh with huggingface_hub
61b81ef verified
raw
history blame
1.03 kB
#!/bin/bash
# LoRA train script by @Akegarasu
config_file="./config/default.toml" # config file | 使用 toml 文件指定训练参数
sample_prompts="./config/sample_prompts.txt" # prompt file for sample | 采样 prompts 文件, 留空则不启用采样功能
sdxl=0 # train sdxl LoRA | 训练 SDXL LoRA
multi_gpu=0 # multi gpu | 多显卡训练 该参数仅限在显卡数 >= 2 使用
# ============= DO NOT MODIFY CONTENTS BELOW | 请勿修改下方内容 =====================
export HF_HOME="huggingface"
export TF_CPP_MIN_LOG_LEVEL=3
export PYTHONUTF8=1
extArgs=()
launchArgs=()
if [[ $multi_gpu == 1 ]]; then
launchArgs+=("--multi_gpu")
launchArgs+=("--num_processes=2")
fi
# run train
if [[ $sdxl == 1 ]]; then
script_name="sdxl_train_network.py"
else
script_name="train_network.py"
fi
python -m accelerate.commands.launch "${launchArgs[@]}" --num_cpu_threads_per_process=8 "./scripts/$script_name" \
--config_file="$config_file" \
--sample_prompts="$sample_prompts" \
"${extArgs[@]}"