| { |
| "architectures": [ |
| "LlamaForCausalLMEagle3" |
| ], |
| "bos_token_id": 199998, |
| "eos_token_id": 200002, |
| "hidden_act": "silu", |
| "hidden_size": 2880, |
| "initializer_range": 0.02, |
| "intermediate_size": 16384, |
| "max_position_embeddings": 131072, |
| "model_type": "llama", |
| "num_attention_heads": 64, |
| "num_key_value_heads": 8, |
| "num_hidden_layers": 1, |
| "pad_token_id": 199999, |
| "rms_norm_eps": 1e-05, |
| "tie_word_embeddings": false, |
| "torch_dtype": "bfloat16", |
| "transformers_version": "4.57.1", |
| "use_cache": true, |
| "vocab_size": 201088, |
| "draft_vocab_size": 201088, |
| "rope_scaling": { |
| "beta_fast": 32.0, |
| "beta_slow": 1.0, |
| "factor": 32.0, |
| "original_max_position_embeddings": 4096, |
| "rope_type": "yarn", |
| "truncate": false |
| }, |
| "attention_bias": false, |
| "attention_dropout": 0.0, |
| "head_dim": 64, |
| "mlp_bias": false, |
| "pretraining_tp": null, |
| "rope_theta": 500000.0, |
| "eagle_config": { |
| "eagle_aux_hidden_state_layer_ids": [], |
| "use_aux_hidden_state": false, |
| "use_input_layernorm_in_first_layer": true, |
| "use_last_layernorm": true, |
| "use_mtp_layernorm": false |
| } |
| } |
