{
  "architectures": [
    "PhimoeForCausalLM"
  ],
  "attention_bias": true,
  "attention_dropout": 0.0,
  "auto_map": {
    "AutoConfig": "configuration_slimmoe.PhiMoEConfig",
    "AutoModelForCausalLM": "modeling_slimmoe.PhiMoEForCausalLM"
  },
  "bos_token_id": 1,
  "dtype": "bfloat16",
  "eos_token_id": 32000,
  "expert_dropout": 0.0,
  "head_dim": 128,
  "hidden_act": "silu",
  "hidden_dropout": 0.0,
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "input_jitter_noise": 0.01,
  "intermediate_size": 960,
  "is_pruned": true,
  "lm_head_bias": true,
  "max_position_embeddings": 4096,
  "model_type": "phimoe",
  "num_attention_heads": 32,
  "num_experts_per_tok": 2,
  "num_hidden_layers": 32,
  "num_key_value_heads": 8,
  "num_local_experts": 48,
  "old_to_new_mapping": {
    "0": {
      "4": 0
    },
    "1": {
      "14": 3,
      "3": 0,
      "4": 1,
      "5": 2
    },
    "10": {
      "1": 0,
      "2": 1,
      "6": 2
    },
    "11": {
      "0": 0,
      "14": 2,
      "6": 1
    },
    "12": {
      "10": 1,
      "12": 2,
      "6": 0
    },
    "13": {
      "11": 2,
      "13": 3,
      "14": 4,
      "2": 0,
      "3": 1
    },
    "14": {
      "2": 0,
      "3": 1
    },
    "15": {
      "1": 0,
      "10": 2,
      "12": 3,
      "14": 4,
      "4": 1
    },
    "16": {
      "1": 0,
      "12": 3,
      "3": 1,
      "5": 2
    },
    "17": {
      "4": 0,
      "7": 1
    },
    "18": {
      "10": 2,
      "14": 3,
      "4": 0,
      "5": 1
    },
    "19": {
      "10": 3,
      "14": 4,
      "4": 0,
      "5": 1,
      "7": 2
    },
    "2": {
      "1": 0,
      "12": 5,
      "14": 6,
      "3": 1,
      "5": 2,
      "6": 3,
      "8": 4
    },
    "20": {
      "1": 0,
      "12": 5,
      "13": 6,
      "14": 7,
      "2": 1,
      "5": 2,
      "6": 3,
      "7": 4
    },
    "21": {
      "14": 1,
      "15": 2,
      "4": 0
    },
    "22": {
      "14": 1,
      "15": 2,
      "3": 0
    },
    "23": {
      "12": 3,
      "14": 4,
      "3": 0,
      "5": 1,
      "9": 2
    },
    "24": {
      "11": 2,
      "5": 0,
      "9": 1
    },
    "25": {
      "0": 0,
      "10": 2,
      "11": 3,
      "9": 1
    },
    "26": {
      "12": 2,
      "2": 0,
      "8": 1
    },
    "27": {
      "10": 3,
      "12": 4,
      "13": 5,
      "15": 6,
      "4": 0,
      "5": 1,
      "9": 2
    },
    "28": {
      "10": 2,
      "11": 3,
      "12": 4,
      "13": 5,
      "4": 0,
      "7": 1
    },
    "29": {
      "1": 0,
      "4": 1,
      "6": 2
    },
    "3": {
      "15": 1,
      "6": 0
    },
    "30": {
      "1": 0,
      "10": 3,
      "11": 4,
      "12": 5,
      "15": 6,
      "3": 1,
      "7": 2
    },
    "31": {
      "10": 1,
      "7": 0
    },
    "4": {
      "10": 3,
      "13": 4,
      "4": 0,
      "7": 1,
      "9": 2
    },
    "5": {},
    "6": {
      "0": 0,
      "1": 1,
      "10": 5,
      "13": 6,
      "6": 2,
      "7": 3,
      "9": 4
    },
    "7": {
      "1": 0,
      "10": 2,
      "13": 3,
      "7": 1
    },
    "8": {
      "10": 3,
      "12": 4,
      "14": 5,
      "3": 0,
      "5": 1,
      "6": 2
    },
    "9": {
      "0": 0,
      "11": 2,
      "4": 1
    }
  },
  "output_router_logits": false,
  "pruned": true,
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "router_aux_loss_coef": 0.0,
  "router_jitter_noise": 0.01,
  "sliding_window": 2047,
  "tie_word_embeddings": false,
  "transformers_version": "4.57.3",
  "use_cache": true,
  "vocab_size": 32064
}