| { |
| "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.28.projections", |
| "backbone.blocks.24.mlp.l1", |
| "backbone.blocks.31.projections", |
| "backbone.blocks.24.inner_mha_cls.Wqkv", |
| "backbone.blocks.26.mlp.l1", |
| "backbone.blocks.27.mlp.l3", |
| "backbone.blocks.28.mlp.l2", |
| "backbone.blocks.31.mlp.l1", |
| "backbone.blocks.29.out_filter_dense", |
| "backbone.blocks.31.mlp.l3", |
| "backbone.blocks.27.out_filter_dense", |
| "backbone.blocks.29.mlp.l2", |
| "backbone.blocks.25.out_filter_dense", |
| "backbone.blocks.31.mlp.l2", |
| "backbone.blocks.29.mlp.l3", |
| "backbone.blocks.27.mlp.l2", |
| "backbone.blocks.25.projections", |
| "backbone.blocks.30.mlp.l3", |
| "backbone.blocks.25.mlp.l1", |
| "backbone.blocks.27.projections", |
| "backbone.blocks.27.mlp.l1", |
| "backbone.blocks.31.out_filter_dense", |
| "backbone.blocks.28.mlp.l3", |
| "backbone.blocks.25.mlp.l2", |
| "backbone.blocks.30.mlp.l1", |
| "backbone.blocks.24.mlp.l3", |
| "backbone.blocks.30.out_filter_dense", |
| "backbone.blocks.24.inner_mha_cls.out_proj", |
| "backbone.blocks.25.mlp.l3", |
| "backbone.blocks.26.projections", |
| "backbone.blocks.24.mlp.l2", |
| "backbone.blocks.26.mlp.l3", |
| "backbone.blocks.28.mlp.l1", |
| "backbone.blocks.26.mlp.l2", |
| "backbone.blocks.29.projections", |
| "backbone.blocks.29.mlp.l1", |
| "backbone.blocks.26.out_filter_dense", |
| "backbone.blocks.28.out_filter_dense", |
| "backbone.blocks.30.projections", |
| "backbone.blocks.30.mlp.l2" |
| ], |
| "task_type": null, |
| "use_dora": false, |
| "use_rslora": false |
| } |