{
  "_name_or_path": "./diff_attention_model",
  "architectures": [
    "DiffAttentionPreTrainedModel"
  ],
  "embed_dim": 768,
  "hidden_dim": 768,
  "lambda_": 0.5,
  "lambda_init": 0.1,
  "max_seq_len": 1024,
  "model_type": "diff_attention_lm",
  "num_heads": 6,
  "num_layers": 12,
  "torch_dtype": "float32",
  "transformers_version": "4.46.2",
  "vocab_size": 50257
}