{
  "architectures": [
    "GPTSanJapaneseForConditionalGeneration"
  ],
  "bos_token_id": 35993,
  "d_ext": 4096,
  "d_ff": 8192,
  "d_model": 1024,
  "d_spout": 128,
  "do_sample": true,
  "dropout_rate": 0.0,
  "eos_token_id": 35999,
  "expert_capacity": 128,
  "initializer_factor": 0.002,
  "layer_norm_epsilon": 1e-05,
  "max_position_embeddings": 1280,
  "model_type": "gptsan-japanese",
  "num_experts": 16,
  "num_ext_layers": 0,
  "num_heads": 16,
  "num_layers": 10,
  "num_switch_layers": 10,
  "output_router_logits": false,
  "pad_token_id": 35995,
  "router_bias": false,
  "router_dtype": "float32",
  "router_ignore_padding_tokens": false,
  "router_jitter_noise": 0.0,
  "separator_token_id": 35998,
  "top_k": 120,
  "torch_dtype": "float32",
  "transformers_version": "4.27.0.dev0",
  "unk_token_id": 35996,
  "use_cache": true,
  "vocab_size": 36000
}