| { | |
| "bf16": { | |
| "enabled": true | |
| }, | |
| "zero_optimization": { | |
| "stage": 1, | |
| "overlap_comm": true, | |
| "contiguous_gradients": true, | |
| "allgather_bucket_size": 200000000, | |
| "reduce_bucket_size": 200000000 | |
| }, | |
| "gradient_clipping": 1.0, | |
| "train_batch_size": "auto", | |
| "train_micro_batch_size_per_gpu": "auto", | |
| "gradient_accumulation_steps": "auto", | |
| "steps_per_print": 10, | |
| "wall_clock_breakdown": false, | |
| "sequence_parallel_size": 8, | |
| "data_efficiency": { | |
| "enabled": false | |
| } | |
| } |