| { | |
| "optimizer_state_offchip": true, | |
| "replicated_tensor_sharding": true, | |
| "enable_half_first_order_momentum": true, | |
| "enable_half_partials": true, | |
| "recompute_checkpoint_every_layer": false, | |
| "device_iterations": 2, | |
| "inference_device_iterations": 4, | |
| "replication_factor": 2, | |
| "inference_replication_factor": 2, | |
| "gradient_accumulation_steps": 16, | |
| "executable_cache_dir": "./exe_cache", | |
| "ipus_per_replica": 8, | |
| "layers_per_ipu": [2, 1, 4, 3, 3, 3, 3, 0], | |
| "matmul_proportion": [0.1, 0.1, 0.1, 0.3, 0.3, 0.3, 0.3, 0.1] | |
| } |