| { | |
| "backbone_checkpoint_name": "t5-base", | |
| "backbone_class": "T5ForConditionalGeneration", | |
| "backbone_hash": "6d029f2b782e1049f4ca25890fc9b113", | |
| "bottleneck_dim": null, | |
| "common_structure": true, | |
| "delta_type": "compacter", | |
| "factorized_phm": true, | |
| "factorized_phm_rule": false, | |
| "hypercomplex_division": 4, | |
| "hypercomplex_nonlinearity": "glorot-uniform", | |
| "kronecker_prod": null, | |
| "learn_phm": true, | |
| "modified_modules": [ | |
| "attn", | |
| "ff" | |
| ], | |
| "non_linearity": "gelu_new", | |
| "opendelta_version": "0.0.1", | |
| "phm_c_init": "normal", | |
| "phm_init_range": 0.0001, | |
| "phm_rank": 1, | |
| "reduction_factor": 16, | |
| "sequential": null, | |
| "shared_W_phm": false, | |
| "shared_phm_rule": false, | |
| "transformers_version": "4.16.0.dev0", | |
| "unfrozen_modules": [ | |
| "deltas", | |
| "layer_norm", | |
| "final_layer_norm" | |
| ], | |
| "use_bias_down_sampler": true, | |
| "use_bias_up_sampler": true | |
| } | |