{
  "architectures": [
    "LlamaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoModelForCausalLM": "modeling_lean_llama.LeanLlamaForCausalLM"
  },
  "bos_token_id": 128000,
  "dtype": "float16",
  "eos_token_id": [
    128001,
    128008,
    128009
  ],
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 14336,
  "leanllm_compressed_kv_layers": [
    18,
    20,
    24,
    26,
    28,
    30
  ],
  "leanllm_kv_decoder_depth": {
    "18": 2,
    "20": 2,
    "24": 2,
    "26": 2,
    "28": 2,
    "30": 2
  },
  "leanllm_kv_decoder_hidden_dim": {
    "18": 256,
    "20": 256,
    "24": 256,
    "26": 256,
    "28": 256,
    "30": 256
  },
  "leanllm_kv_fixed_alpha": {
    "18": 1.0,
    "20": 1.0,
    "24": 1.0,
    "26": 1.0,
    "28": 1.0,
    "30": 1.0
  },
  "leanllm_kv_granularity": {
    "18": "per_token",
    "20": "per_token",
    "24": "per_token",
    "26": "per_token",
    "28": "per_token",
    "30": "per_token"
  },
  "leanllm_kv_input_dims": {
    "18": 1024,
    "20": 1024,
    "24": 1024,
    "26": 1024,
    "28": 1024,
    "30": 1024
  },
  "leanllm_kv_key_dims": {
    "18": 0,
    "20": 0,
    "24": 0,
    "26": 0,
    "28": 0,
    "30": 0
  },
  "leanllm_kv_residual_decoder": {
    "18": false,
    "20": false,
    "24": false,
    "26": false,
    "28": false,
    "30": false
  },
  "leanllm_kv_value_dims": {
    "18": 4,
    "20": 4,
    "24": 4,
    "26": 4,
    "28": 4,
    "30": 4
  },
  "leanllm_values_only": true,
  "max_position_embeddings": 131072,
  "mlp_bias": false,
  "model_type": "llama",
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "num_key_value_heads": 8,
  "pad_token_id": null,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_parameters": {
    "factor": 8.0,
    "high_freq_factor": 4.0,
    "low_freq_factor": 1.0,
    "original_max_position_embeddings": 8192,
    "rope_theta": 500000.0,
    "rope_type": "llama3"
  },
  "tie_word_embeddings": false,
  "transformers_version": "5.1.0",
  "use_cache": true,
  "vocab_size": 128256
}