| { | |
| "zero_optimization": { | |
| "stage": 2, | |
| "offload_optimizer": { | |
| "device": "cpu", | |
| "pin_memory": true | |
| }, | |
| "allgather_partitions": true, | |
| "allgather_bucket_size": 2e8, | |
| "overlap_comm": true, | |
| "reduce_scatter": true, | |
| "reduce_bucket_size": 2e8, | |
| "contiguous_gradients": true | |
| }, | |
| "activation_checkpointing": { | |
| "partition_activations": true, | |
| "contiguous_memory_optimization": true, | |
| "cpu_checkpointing": true, | |
| "number_checkpoints": 2 | |
| }, | |
| "bf16": { | |
| "enabled": true | |
| }, | |
| "optimizer": { | |
| "type": "AdamW", | |
| "params": { | |
| "lr": "auto", | |
| "betas": [0.9, 0.999], | |
| "eps": 1e-8, | |
| "weight_decay": "auto" | |
| } | |
| }, | |
| "gradient_clipping": 1.0, | |
| "train_micro_batch_size_per_gpu": 1, | |
| "train_batch_size": 4, | |
| "steps_per_print": 1, | |
| "wall_clock_breakdown": false, | |
| "zero_allow_untested_optimizer": true, | |
| "zero_force_ds_cpu_optimizer": false, | |
| "dump_state": true, | |
| "verbose": true, | |
| "gradient_accumulation_steps": 2 | |
| } |