| { |
| "architectures": [ |
| "Evo2ForCausalLM" |
| ], |
| "attn_dropout": 0.0, |
| "dtype": "float32", |
| "eos_token_id": 0, |
| "hidden_dropout": 0.0, |
| "hidden_size": 1920, |
| "hyena_filter_configurations": [ |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| {}, |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| {}, |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| {}, |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| { |
| "h_shape": [ |
| 128, |
| 1, |
| 7 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "h_shape": [ |
| 128, |
| 1, |
| 128 |
| ] |
| }, |
| { |
| "D_shape": [ |
| 1920 |
| ], |
| "log_poles_shape": [ |
| 1920, |
| 16, |
| 1 |
| ], |
| "residues_shape": [ |
| 1920, |
| 16 |
| ] |
| }, |
| {} |
| ], |
| "hyena_filters": 128, |
| "hyena_flip_x1x2": false, |
| "hyena_hidden_size": 1920, |
| "hyena_kernel_size": 3, |
| "hyena_order": 3, |
| "initializer_range": 0.02, |
| "intermediate_size": 5120, |
| "layer_types": [ |
| "hyena", |
| "hyena", |
| "hyena", |
| "attention", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "attention", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "attention", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "hyena", |
| "attention" |
| ], |
| "max_position_embeddings": 2048, |
| "mlp_dropout": 0.0, |
| "model_type": "evo2", |
| "num_attention_heads": 15, |
| "num_hidden_layers": 25, |
| "num_key_value_heads": 15, |
| "pad_token_id": 1, |
| "rms_norm_eps": 1e-06, |
| "rope_parameters": { |
| "rope_theta": 1000000.0, |
| "rope_type": "default" |
| }, |
| "rope_theta": 1000000.0, |
| "transformers_version": "5.0.0.dev0", |
| "use_cache": true, |
| "vocab_size": 512 |
| } |
|
|
