bf16: true
cutoff_len: 4096
dataset: merger,LLM_dataset(4o),LLM_dataset(4mini)
dataset_dir: /media/dl/7DC4-B1CE/500_video
ddp_timeout: 180000000
deepspeed: cache/ds_z2_config.json
do_train: true
finetuning_type: lora
flash_attn: fa2
gradient_accumulation_steps: 8
learning_rate: 5.0e-05
logging_steps: 5
lora_alpha: 16
lora_dropout: 0
lora_rank: 8
lora_target: all
lr_scheduler_type: cosine
max_grad_norm: 1.0
max_samples: 100000
model_name_or_path: llava-hf/LLaVA-NeXT-Video-7B-hf
num_train_epochs: 1.0
optim: adamw_torch
output_dir: saves/LLaVA-NeXT-Video-7B-Chat/lora/train_2024-12-01-18-22-24
packing: false
per_device_train_batch_size: 2
plot_loss: true
preprocessing_num_workers: 16
quantization_bit: 4
quantization_method: bitsandbytes
report_to: none
save_steps: 100
stage: sft
template: llava_next_video
warmup_steps: 100