| { | |
| "auto_wrap_policy": "TRANSFORMER_BASED_WRAP", | |
| "backward_prefetch_policy": "BACKWARD_PRE", | |
| "cpu_ram_efficient_loading": "true", | |
| "forward_prefetch": "false", | |
| "offload_params": "false", | |
| "sharding_strategy": "SHARD_GRAD_OP", | |
| "state_dict_type": "SHARDED_STATE_DICT", | |
| "sync_module_states": "true", | |
| "transformer_layer_cls_to_wrap": "GenerannoEncoderLayer", | |
| "use_orig_params": "true" | |
| } |