jd_train/ds_config_zero3.json

23 lines
691 B
JSON

{
"train_micro_batch_size_per_gpu": 1,
"gradient_accumulation_steps": 7,
"zero_optimization": {
"stage": 3,
"overlap_comm": true,
"contiguous_gradients": true,
"reduce_bucket_size": 200000000,
"stage3_prefetch_bucket_size": 100000000,
"stage3_param_persistence_threshold": 1000000,
"offload_optimizer": { "device": "cpu", "pin_memory": true },
"offload_param": { "device": "none" }
},
"fp16": { "enabled": false },
"bf16": { "enabled": true },
"optimizer": {
"type": "CPUAdam",
"params": { "lr": 2e-5, "betas": [0.9, 0.999], "eps": 1e-8, "weight_decay": 0.1 }
},
"gradient_clipping": 1.0,
"zero_force_ds_cpu_optimizer": true
}