config.json

{
  "_name_or_path": "./llama_deepseek_8B_mla_moe",
  "architectures": [
    "LlamaDeepSeekForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 14336,
  "latent_size": 1024,
  "max_position_embeddings": 131072,
  "mla_layers": [
    8,
    10,
    12,
    14,
    16,
    18,
    20,
    22,
    24
  ],
  "mlp_bias": false,
  "model_type": "llama_deepseek",
  "moe_layers": [
    11,
    15,
    19,
    23
  ],
  "num_attention_heads": 32,
  "num_experts": 8,
  "num_experts_per_tok": 2,
  "num_hidden_layers": 32,
  "num_key_value_heads": 8,
  "num_latent_heads": 16,
  "pretraining_tp": 1,
  "rms_norm_eps": 1e-05,
  "rope_scaling": {
    "factor": 8.0,
    "high_freq_factor": 4.0,
    "low_freq_factor": 1.0,
    "original_max_position_embeddings": 8192,
    "rope_type": "llama3"
  },
  "rope_theta": 500000.0,
  "tie_word_embeddings": false,
  "torch_dtype": "float16",
  "transformers_version": "4.49.0.dev0",
  "use_cache": true,
  "use_mla": true,
  "use_moe": true,
  "vocab_size": 128256
}