qwerrwe / ds_config.json
winglian's picture
8bit and deepspeed changes
9190ada
raw
history blame
1.33 kB
{
"bf16": {
"enabled": "auto"
},
"fp16": {
"enabled": "auto",
"loss_scale": 0,
"loss_scale_window": 1000,
"initial_scale_power": 16,
"hysteresis": 2,
"min_loss_scale": 1
},
"optimizer": {
"type": "Adam",
"params": {
"lr": "auto",
"betas": "auto",
"eps": "auto",
"weight_decay": "auto"
}
},
"scheduler": {
"type": "WarmupDecayLR",
"params": {
"warmup_min_lr": "auto",
"warmup_max_lr": "auto",
"warmup_num_steps": "auto",
"total_num_steps": "auto"
}
},
"zero_optimization": {
"stage": 2,
"offload_optimizer": {
"device": "cpu",
"pin_memory": true
},
"offload_param": {
"device": "cpu",
"pin_memory": true
},
"overlap_comm": true,
"allgather_partitions": true,
"allgather_bucket_size": 5e8,
"contiguous_gradients": true,
"reduce_bucket_size": "auto",
"reduce_scatter": true,
"stage3_max_live_parameters": 0,
"stage3_max_reuse_distance": 0,
"stage3_gather_16bit_weights_on_model_save": true
},
"gradient_accumulation_steps": "auto",
"gradient_clipping": "auto",
"steps_per_print": 5,
"train_batch_size": "auto",
"train_micro_batch_size_per_gpu": "auto",
"wall_clock_breakdown": false,
"round_robin_gradients": true
}