{ | |
"embedding_serialization_factor": 5, | |
"recompute_checkpoint_every_layer": true, | |
"optimizer_state_offchip": true, | |
"replicated_tensor_sharding": true, | |
"enable_half_partials": true, | |
"device_iterations": 1, | |
"inference_device_iterations": 5, | |
"replication_factor": 1, | |
"inference_replication_factor": 1, | |
"gradient_accumulation_steps": 16, | |
"executable_cache_dir": "./exe_cache", | |
"ipus_per_replica": 4, | |
"layers_per_ipu": [3, 7, 7, 7], | |
"matmul_proportion": [0.1, 0.2, 0.25, 0.25] | |
} |