| { |
| "alpha_pattern": {}, |
| "auto_mapping": { |
| "base_model_class": "StripedHyenaModelForCausalLM", |
| "parent_library": "transformers_modules.togethercomputer.evo-1-131k-base.567369e9825aa08b3de4b122fca34fac6a890602.modeling_hyena" |
| }, |
| "base_model_name_or_path": "togethercomputer/evo-1-8k-base", |
| "bias": "none", |
| "fan_in_fan_out": false, |
| "inference_mode": true, |
| "init_lora_weights": "gaussian", |
| "layer_replication": null, |
| "layers_pattern": null, |
| "layers_to_transform": null, |
| "loftq_config": {}, |
| "lora_alpha": 32, |
| "lora_dropout": 0.05, |
| "megatron_config": null, |
| "megatron_core": "megatron.core", |
| "modules_to_save": null, |
| "peft_type": "LORA", |
| "r": 16, |
| "rank_pattern": {}, |
| "revision": null, |
| "target_modules": [ |
| "backbone.blocks.11.projections", |
| "backbone.blocks.2.mlp.l1", |
| "backbone.blocks.21.mlp.l3", |
| "backbone.blocks.0.mlp.l3", |
| "backbone.blocks.18.mlp.l1", |
| "backbone.blocks.26.mlp.l1", |
| "backbone.blocks.22.mlp.l1", |
| "backbone.blocks.26.mlp.l2", |
| "backbone.blocks.31.out_filter_dense", |
| "backbone.blocks.8.inner_mha_cls.out_proj", |
| "backbone.blocks.24.mlp.l1", |
| "backbone.blocks.3.projections", |
| "backbone.blocks.2.projections", |
| "backbone.blocks.7.mlp.l3", |
| "backbone.blocks.27.projections", |
| "backbone.blocks.12.mlp.l3", |
| "backbone.blocks.13.mlp.l1", |
| "backbone.blocks.22.mlp.l2", |
| "backbone.blocks.25.out_filter_dense", |
| "backbone.blocks.29.mlp.l1", |
| "backbone.blocks.28.projections", |
| "backbone.blocks.13.mlp.l3", |
| "backbone.blocks.0.out_filter_dense", |
| "backbone.blocks.12.projections", |
| "backbone.blocks.14.mlp.l1", |
| "backbone.blocks.16.mlp.l1", |
| "backbone.blocks.25.mlp.l3", |
| "backbone.blocks.8.inner_mha_cls.Wqkv", |
| "backbone.blocks.5.mlp.l1", |
| "backbone.blocks.30.mlp.l1", |
| "backbone.blocks.24.mlp.l3", |
| "backbone.blocks.7.mlp.l1", |
| "backbone.blocks.2.mlp.l3", |
| "backbone.blocks.28.mlp.l1", |
| "backbone.blocks.20.mlp.l1", |
| "backbone.blocks.29.projections", |
| "backbone.blocks.12.mlp.l1", |
| "backbone.blocks.5.out_filter_dense", |
| "backbone.blocks.26.mlp.l3", |
| "backbone.blocks.6.mlp.l1", |
| "backbone.blocks.19.mlp.l1", |
| "backbone.blocks.25.mlp.l2", |
| "backbone.blocks.15.projections", |
| "backbone.blocks.23.mlp.l3", |
| "backbone.blocks.5.mlp.l3", |
| "backbone.blocks.30.mlp.l2", |
| "backbone.blocks.1.mlp.l1", |
| "backbone.blocks.18.mlp.l2", |
| "backbone.blocks.10.mlp.l3", |
| "backbone.blocks.27.mlp.l3", |
| "backbone.blocks.27.mlp.l1", |
| "backbone.blocks.20.projections", |
| "backbone.blocks.20.mlp.l2", |
| "backbone.blocks.25.projections", |
| "backbone.blocks.0.mlp.l2", |
| "backbone.blocks.20.out_filter_dense", |
| "backbone.blocks.19.projections", |
| "backbone.blocks.9.mlp.l2", |
| "backbone.blocks.23.mlp.l1", |
| "backbone.blocks.19.mlp.l2", |
| "backbone.blocks.31.mlp.l2", |
| "backbone.blocks.14.mlp.l2", |
| "backbone.blocks.1.projections", |
| "backbone.blocks.1.out_filter_dense", |
| "backbone.blocks.16.mlp.l2", |
| "backbone.blocks.4.out_filter_dense", |
| "backbone.blocks.9.mlp.l3", |
| "backbone.blocks.13.projections", |
| "backbone.blocks.6.out_filter_dense", |
| "backbone.blocks.8.mlp.l2", |
| "backbone.blocks.3.mlp.l2", |
| "backbone.blocks.5.mlp.l2", |
| "backbone.blocks.7.out_filter_dense", |
| "backbone.blocks.3.mlp.l3", |
| "backbone.blocks.4.projections", |
| "backbone.blocks.4.mlp.l1", |
| "backbone.blocks.9.out_filter_dense", |
| "backbone.blocks.13.mlp.l2", |
| "backbone.blocks.23.mlp.l2", |
| "backbone.blocks.10.projections", |
| "backbone.blocks.17.mlp.l1", |
| "backbone.blocks.30.out_filter_dense", |
| "backbone.blocks.31.mlp.l3", |
| "backbone.blocks.29.mlp.l2", |
| "backbone.blocks.11.mlp.l2", |
| "backbone.blocks.24.inner_mha_cls.out_proj", |
| "backbone.blocks.29.mlp.l3", |
| "backbone.blocks.12.mlp.l2", |
| "backbone.blocks.28.mlp.l2", |
| "backbone.blocks.10.out_filter_dense", |
| "backbone.blocks.23.out_filter_dense", |
| "backbone.blocks.11.mlp.l1", |
| "backbone.blocks.28.mlp.l3", |
| "backbone.blocks.18.projections", |
| "backbone.blocks.20.mlp.l3", |
| "backbone.blocks.22.mlp.l3", |
| "backbone.blocks.21.mlp.l1", |
| "backbone.blocks.22.projections", |
| "backbone.blocks.25.mlp.l1", |
| "backbone.blocks.15.mlp.l1", |
| "backbone.blocks.1.mlp.l2", |
| "backbone.blocks.4.mlp.l3", |
| "backbone.blocks.21.mlp.l2", |
| "backbone.blocks.0.mlp.l1", |
| "backbone.blocks.21.out_filter_dense", |
| "backbone.blocks.9.mlp.l1", |
| "backbone.blocks.11.mlp.l3", |
| "backbone.blocks.8.mlp.l3", |
| "backbone.blocks.16.inner_mha_cls.Wqkv", |
| "backbone.blocks.2.mlp.l2", |
| "backbone.blocks.29.out_filter_dense", |
| "backbone.blocks.1.mlp.l3", |
| "backbone.blocks.17.projections", |
| "backbone.blocks.7.projections", |
| "backbone.blocks.19.mlp.l3", |
| "backbone.blocks.0.projections", |
| "backbone.blocks.15.out_filter_dense", |
| "backbone.blocks.15.mlp.l3", |
| "backbone.blocks.24.inner_mha_cls.Wqkv", |
| "backbone.blocks.27.mlp.l2", |
| "backbone.blocks.6.mlp.l3", |
| "backbone.blocks.4.mlp.l2", |
| "backbone.blocks.27.out_filter_dense", |
| "backbone.blocks.15.mlp.l2", |
| "backbone.blocks.31.projections", |
| "backbone.blocks.9.projections", |
| "backbone.blocks.5.projections", |
| "backbone.blocks.11.out_filter_dense", |
| "backbone.blocks.7.mlp.l2", |
| "backbone.blocks.22.out_filter_dense", |
| "backbone.blocks.8.mlp.l1", |
| "backbone.blocks.10.mlp.l1", |
| "backbone.blocks.21.projections", |
| "backbone.blocks.17.mlp.l2", |
| "backbone.blocks.30.projections", |
| "backbone.blocks.6.mlp.l2", |
| "backbone.blocks.18.mlp.l3", |
| "backbone.blocks.23.projections", |
| "backbone.blocks.13.out_filter_dense", |
| "backbone.blocks.26.projections", |
| "backbone.blocks.14.projections", |
| "backbone.blocks.31.mlp.l1", |
| "backbone.blocks.3.mlp.l1", |
| "backbone.blocks.18.out_filter_dense", |
| "backbone.blocks.2.out_filter_dense", |
| "backbone.blocks.12.out_filter_dense", |
| "backbone.blocks.28.out_filter_dense", |
| "backbone.blocks.10.mlp.l2", |
| "backbone.blocks.24.mlp.l2", |
| "backbone.blocks.17.out_filter_dense", |
| "backbone.blocks.17.mlp.l3", |
| "backbone.blocks.19.out_filter_dense", |
| "backbone.blocks.14.mlp.l3", |
| "backbone.blocks.30.mlp.l3", |
| "backbone.blocks.16.mlp.l3", |
| "backbone.blocks.16.inner_mha_cls.out_proj", |
| "backbone.blocks.14.out_filter_dense", |
| "backbone.blocks.26.out_filter_dense", |
| "backbone.blocks.6.projections", |
| "backbone.blocks.3.out_filter_dense" |
| ], |
| "task_type": null, |
| "use_dora": false, |
| "use_rslora": false |
| } |