{
  "architectures": [
    "BertModel"
  ],
  "model_type": "bert",
  "attention_probs_dropout_prob": 0.1,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "initializer_range": 0.02,
  "num_attention_heads": 12,
  "pad_token_id": 0,
  "position_embedding_type": "absolute",
  "tokenizer_class": "XLMRobertaTokenizer",
  "vocab_size": 250000,
  "hidden_size": 384,
  "num_hidden_layers": 12,
  "intermediate_size": 1536,
  "max_position_embeddings": 512,
  "type_vocab_size": 2,
  "layer_norm_eps": 1e-12,
  "ultracode": {
    "format": "raw-safetensors/1",
    "exported_from": "model.onnx",
    "onnx_sha256": "3d527785e74b24fa53f3c04dccd5e71f95288411b523be013052181acce86c9c",
    "head": "mean_l2",
    "position_ids": "absolute",
    "pooling": "mean over attention_mask",
    "normalize": "l2",
    "vocab_remap_folded": {
      "input_ids": 250000,
      "pruned_rows": 142076
    },
    "input_ids": "raw SentencePiece ids (no fairseq offset), as the graph takes them"
  }
}