bf16: true
cutoff_len: 1024
dataset: mathinstruct
dataset_dir: data
ddp_timeout: 180000000
do_train: true
finetuning_type: lora
flash_attn: auto
gradient_accumulation_steps: 8
include_num_input_tokens_seen: true
learning_rate: 5.0e-05
logging_steps: 5
lora_alpha: 2
lora_dropout: 0
lora_rank: 1
lora_target: all
lr_scheduler_type: cosine
max_grad_norm: 1.0
max_samples: 100000
model_name_or_path: D:/models/Qwen2.5-0.5B
num_train_epochs: 1.0
optim: adamw_torch
output_dir: saves\Qwen2.5-0.5B\lora\train_2024-09-26-14-50-59
packing: false
per_device_train_batch_size: 2
plot_loss: true
preprocessing_num_workers: 16
quantization_bit: 4
quantization_method: bitsandbytes
report_to: all
save_steps: 100
stage: sft
template: qwen
warmup_steps: 0