mirror of
https://github.com/modelscope/modelscope.git
synced 2026-02-24 20:19:51 +01:00
f4c90f2adf
Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/13080086 * support lora for llama * update baichuan * remove work_dir * fixbug: 1. change ConfigDict to list when hooks key not in config 2. ignore all bin files when preparing output folder * 1. support device_map 2. remove the operation of to float when using lora * add inference file * add comment * support device_map
45 lines
1.1 KiB
JSON
45 lines
1.1 KiB
JSON
{
|
|
"bf16": {
|
|
"enabled": "auto"
|
|
},
|
|
"optimizer": {
|
|
"type": "AdamW",
|
|
"params": {
|
|
"lr": "auto",
|
|
"betas": "auto",
|
|
"eps": "auto",
|
|
"weight_decay": "auto"
|
|
}
|
|
},
|
|
"scheduler": {
|
|
"type": "WarmupDecayLR",
|
|
"params": {
|
|
"total_num_steps": "auto",
|
|
"warmup_min_lr": "auto",
|
|
"warmup_max_lr": "auto",
|
|
"warmup_num_steps": "auto"
|
|
}
|
|
},
|
|
"zero_optimization": {
|
|
"stage": 2,
|
|
"offload_optimizer": {
|
|
"device": "cpu",
|
|
"pin_memory": true
|
|
},
|
|
"offload_param": {
|
|
"device": "cpu",
|
|
"pin_memory": true
|
|
},
|
|
"overlap_comm": true,
|
|
"contiguous_gradients": true,
|
|
"sub_group_size": 1e9,
|
|
"reduce_bucket_size": "auto"
|
|
},
|
|
"gradient_accumulation_steps": "auto",
|
|
"gradient_clipping": "auto",
|
|
"steps_per_print": 10000000,
|
|
"train_batch_size": "auto",
|
|
"train_micro_batch_size_per_gpu": "auto",
|
|
"wall_clock_breakdown": false
|
|
}
|