{
  "answer_digits": 5,
  "architecture_version": 2,
  "architectures": [
    "AdditionForCausalLM"
  ],
  "attention_logit_divisor": "d_head",
  "attention_output_scale": 0.3333333333333333,
  "auto_map": {
    "AutoConfig": "configuration_addition.AdditionConfig",
    "AutoModelForCausalLM": "modeling_addition.AdditionForCausalLM"
  },
  "base": 10,
  "bias": false,
  "bos_token_id": 0,
  "d_mlp": 128,
  "d_model": 64,
  "digit_token_offset": 3,
  "dtype": "float32",
  "eos_token_id": null,
  "equals_token_id": 2,
  "export_format_version": 2,
  "hidden_size": 64,
  "init_mode": "normal",
  "intermediate_size": 128,
  "is_decoder": true,
  "max_position_embeddings": 16,
  "max_seq_len": 16,
  "model_type": "fixed-width-addition",
  "n_heads": 4,
  "n_layers": 1,
  "normalization": "none",
  "num_attention_heads": 4,
  "num_hidden_layers": 1,
  "operand_digits": 4,
  "pad_token_id": null,
  "plus_token_id": 1,
  "residual_alpha": 0.5,
  "tie_word_embeddings": false,
  "transformers_version": "5.15.0",
  "use_cache": false,
  "vocab_size": 13
}