config.json

{
  "architectures": [
    "Eagle3Speculator"
  ],
  "auto_map": {
    "": "eagle3.Eagle3SpeculatorConfig"
  },
  "draft_vocab_size": 32000,
  "has_no_defaults_at_init": false,
  "norm_before_residual": true,
  "speculators_config": {
    "algorithm": "eagle3",
    "default_proposal_method": "greedy",
    "proposal_methods": [
      {
        "accept_tolerance": 0.0,
        "proposal_type": "greedy",
        "speculative_tokens": 3,
        "verifier_accept_k": 1
      }
    ],
    "verifier": {
      "architectures": [
        "LlamaForCausalLM"
      ],
      "name_or_path": "meta-llama/Llama-3.3-70B-Instruct"
    }
  },
  "speculators_model_type": "eagle3",
  "speculators_version": "0.2.0.dev8",
  "target_hidden_size": null,
  "torch_dtype": "bfloat16",
  "transformer_layer_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 8192,
    "initializer_range": 0.02,
    "intermediate_size": 28672,
    "max_position_embeddings": 131072,
    "mlp_bias": false,
    "model_type": "llama",
    "num_attention_heads": 64,
    "num_hidden_layers": 1,
    "num_key_value_heads": 8,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-05,
    "rope_scaling": null,
    "rope_theta": 500000.0,
    "torch_dtype": "bfloat16",
    "use_cache": true,
    "vocab_size": 128256
  },
  "transformers_version": "4.53.2"
}