{ | |
"device_iterations": 1, | |
"embedding_serialization_factor": 2, | |
"enable_half_first_order_momentum": true, | |
"enable_half_partials": true, | |
"executable_cache_dir": "./exe_cache", | |
"gradient_accumulation_steps": 512, | |
"inference_device_iterations": 4, | |
"inference_replication_factor": 4, | |
"ipus_per_replica": 4, | |
"layers_per_ipu": [ | |
0, | |
4, | |
4, | |
4 | |
], | |
"matmul_proportion": 0.22, | |
"optimizer_state_offchip": false, | |
"optimum_version": "0.1.2", | |
"output_mode": "final", | |
"profile_dir": null, | |
"recompute_checkpoint_every_layer": true, | |
"replicated_tensor_sharding": true, | |
"replication_factor": 4, | |
"seed": 42, | |
"use_popdist": false | |
} | |