config.json

{
  "_name_or_path": "jargon-legal-4096",
  "activation": "relu",
  "add_bias_kv": false,
  "add_pooling_layer": false,
  "add_zero_attn": false,
  "architectures": [
    "JargonForMaskedLM"
  ],
  "attention_probs_dropout_prob": 0.1,
  "auto_map": {
    "AutoConfig": "jargon_configuration.JargonConfig",
    "AutoModel": "jargon_model.JargonModel",
    "AutoModelForMaskedLM": "jargon_model.JargonForMaskedLM",
    "AutoModelForSequenceClassification": "jargon_model.JargonForSequenceClassification",
    "AutoModelForTokenClassification": "jargon_model.JargonForTokenClassification"
  },
  "bias": true,
  "bos_token_id": 0,
  "classifier_dropout": null,
  "compress_layer": 1,
  "compressed": 16,
  "dim_feedforward": 4096,
  "dropout": 0.1,
  "embed_dim": 768,
  "encoder_attention_heads": 12,
  "encoder_decoder_attention": false,
  "encoder_embed_dim": 768,
  "encoder_ffn_embed_dim": 4096,
  "encoder_normalize_before": true,
  "eos_token_id": 2,
  "freeze_compress": 0,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.1,
  "hidden_size": 768,
  "initializer_range": 0.02,
  "intermediate_act_fn": "gelu",
  "intermediate_size": 4096,
  "layer_norm_eps": 1e-05,
  "layernorm_embedding": false,
  "max_position_embeddings": 4098,
  "max_positions": 4096,
  "model_type": "jargon",
  "num_attention_heads": 12,
  "num_heads": 16,
  "num_hidden_layers": 12,
  "num_layers": 12,
  "pad_token_id": 1,
  "position_embedding_type": "learned",
  "q_noise": 0,
  "qn_block_size": 8,
  "quant_noise_pq": 0.0,
  "quant_noise_pq_block_size": 8,
  "quant_noise_scalar": 0,
  "self_attention": true,
  "shared_kv_compressed": 1,
  "shared_layer_kv_compressed": 1,
  "torch_dtype": "float32",
  "transformers_version": "4.31.0",
  "type_vocab_size": 2,
  "untie_weights_roberta": false,
  "use_cache": true,
  "vocab_size": 50000
}