config.json

{
    "architectures": [
        "BARTModel",
        "TransformerModel"
    ],
    "task": "translation",
    "source_lang": "en",
    "target_lang": "en",
    "dropout": 0.1,
    "attention_dropout": 0.1,
    "activation_dropout": 0.0,
    "encoder_layers": 12,
    "decoder_layers": 12,
    "encoder_embed_dim": 1024,
    "decoder_embed_dim": 1024,
    "encoder_ffn_embed_dim": 4096,
    "decoder_ffn_embed_dim": 4096,
    "encoder_attention_heads": 16,
    "decoder_attention_heads": 16,
    "encoder_normalize_before": false,
    "decoder_normalize_before": false,
    "encoder_learned_pos": true,
    "decoder_learned_pos": true,
    "share_decoder_input_output_embed": true,
    "share_all_embeddings": true,
    "static_position_embeddings": false,
    "adaptive_input": false,
    "adaptive_input_factor": 4,
    "adaptive_input_cutoff": null,
    "no_scale_embedding": true,
    "layernorm_embedding": true,
    "tie_adaptive_weights": false,
    "tie_encoder_decoder": false,
    "remove_bpe": null,
    "eos_token": "</s>",
    "bos_token": "<s>",
    "pad_token": "<pad>",
    "unk_token": "<unk>",
    "mask_token": "<mask>",
    "layernorm_eps": 1e-05,
    "transformers_version": "4.14.0",
    "fairseq_version": "0.10.4",
    "model_type": "bart",
    "revision": "main",
    "use_cache": true
}