vikp/donut-decoder

1{2  "activation_dropout": 0.0,3  "activation_function": "gelu",4  "add_cross_attention": true,5  "add_final_layer_norm": true,6  "architectures": [7    "MBartForCausalLM"8  ],9  "attention_dropout": 0.0,10  "bos_token_id": 0,11  "classifier_dropout": 0.0,12  "d_model": 1024,13  "decoder_attention_heads": 16,14  "decoder_ffn_dim": 4096,15  "decoder_layerdrop": 0.0,16  "decoder_layers": 4,17  "dropout": 0.1,18  "encoder_attention_heads": 16,19  "encoder_ffn_dim": 4096,20  "encoder_layerdrop": 0.0,21  "encoder_layers": 12,22  "eos_token_id": 2,23  "forced_eos_token_id": 2,24  "init_std": 0.02,25  "is_decoder": true,26  "is_encoder_decoder": false,27  "max_position_embeddings": 1536,28  "model_type": "mbart",29  "num_hidden_layers": 12,30  "pad_token_id": 1,31  "scale_embedding": true,32  "torch_dtype": "float32",33  "transformers_version": "4.36.2",34  "use_cache": true,35  "vocab_size": 5752536}37