Fun-Audio-Chat/training/configs/sft.yaml at main · FunAudioLLM/Fun-Audio-Chat · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
### model
model_name_or_path: ../pretrained_models/Fun-Audio-Chat-8B

### method
stage: sft
do_train: true
do_eval: true
finetuning_type: full
flash_attn: fa2
freeze_vision_tower: false
freeze_multi_modal_projector: false
deepspeed: configs/deepspeed/ds_z2_config.json
print_param_status: true

### dataset
dataset: spoken-alpaca-gpt4_1k
template: funaudiochat
cutoff_len: 2048
overwrite_cache: true
preprocessing_num_workers: 32
preprocessing_batch_size: 10000
neat_packing: true

### output
output_dir: saves/Fun-Audio-Chat-8B/SFT25122101
logging_steps: 10
save_strategy: steps
save_steps: 1000
plot_loss: true
overwrite_output_dir: true

### train
per_device_train_batch_size: 6
gradient_accumulation_steps: 1
learning_rate: 1.0e-5
num_train_epochs: 1.0
lr_scheduler_type: cosine_with_min_lr
lr_scheduler_kwargs: {min_lr_rate: 0.1}
warmup_ratio: 0.02
bf16: true
report_to: none
ddp_timeout: 180000000
disable_gradient_checkpointing: false

### eval
val_size: 0.01
per_device_eval_batch_size: 6
eval_strategy: steps
eval_steps: 1000