config.json

{
  "activation_dropout": 0.0,
  "activation_function": "swish",
  "architectures": [
    "MarianMTModel"
  ],
  "attention_dropout": 0.0,
  "bad_words_ids": [
    [
      96000
    ]
  ],
  "bos_token_id": 0,
  "classifier_dropout": 0.0,
  "d_model": 512,
  "decoder_attention_heads": 8,
  "decoder_ffn_dim": 2048,
  "decoder_layerdrop": 0.0,
  "decoder_layers": 6,
  "decoder_start_token_id": 96000,
  "dropout": 0.1,
  "encoder_attention_heads": 8,
  "encoder_ffn_dim": 2048,
  "encoder_layerdrop": 0.0,
  "encoder_layers": 6,
  "eos_token_id": 0,
  "forced_eos_token_id": 0,
  "gradient_checkpointing": false,
  "init_std": 0.02,
  "is_encoder_decoder": true,
  "max_length": 512,
  "max_position_embeddings": 512,
  "model_type": "marian",
  "normalize_embedding": false,
  "num_beams": 12,
  "num_hidden_layers": 6,
  "pad_token_id": 96000,
  "scale_embedding": true,
  "static_position_embeddings": true,
  "torch_dtype": "float16",
  "transformers_version": "4.9.2",
  "use_cache": true,
  "vocab_size": 96001
}