{ | |
"device_iterations": 1, | |
"embedding_serialization_factor": 0, | |
"enable_half_first_order_momentum": true, | |
"enable_half_partials": true, | |
"executable_cache_dir": "./exe_cache", | |
"gradient_accumulation_steps": 32, | |
"inference_device_iterations": 4, | |
"inference_replication_factor": 4, | |
"ipus_per_replica": 4, | |
"layers_per_ipu": [ | |
0, | |
4, | |
4, | |
4 | |
], | |
"matmul_proportion": [ | |
0.1, | |
0.1, | |
0.1, | |
0.1 | |
], | |
"optimizer_state_offchip": true, | |
"optimum_version": "1.0.0", | |
"output_mode": "final", | |
"profile_dir": "", | |
"recompute_checkpoint_every_layer": true, | |
"replicated_tensor_sharding": true, | |
"replication_factor": 4, | |
"seed": 1984, | |
"transformers_version": "4.18.0.dev0", | |
"use_popdist": false | |
} | |