hf-tiny-v2/tiny-random-EncoderDecoderModel-bert-bert

1{2  "architectures": [3    "EncoderDecoderModel"4  ],5  "decoder": {6    "_name_or_path": "/tmp/tmpk7yofmau/decoder/BertLMHeadModel",7    "add_cross_attention": true,8    "architectures": [9      "BertLMHeadModel"10    ],11    "attention_probs_dropout_prob": 0.1,12    "bos_token_id": null,13    "chunk_size_feed_forward": 0,14    "classifier_dropout": null,15    "dtype": "float32",16    "eos_token_id": null,17    "hidden_act": "gelu",18    "hidden_dropout_prob": 0.1,19    "hidden_size": 32,20    "id2label": {21      "0": "LABEL_0",22      "1": "LABEL_1"23    },24    "initializer_range": 0.02,25    "intermediate_size": 37,26    "is_decoder": true,27    "is_encoder_decoder": false,28    "label2id": {29      "LABEL_0": 0,30      "LABEL_1": 131    },32    "layer_norm_eps": 1e-12,33    "max_position_embeddings": 512,34    "model_type": "bert",35    "num_attention_heads": 4,36    "num_hidden_layers": 2,37    "output_attentions": false,38    "output_hidden_states": false,39    "pad_token_id": 0,40    "problem_type": null,41    "return_dict": true,42    "tie_word_embeddings": true,43    "type_vocab_size": 16,44    "use_cache": true,45    "vocab_size": 3052246  },47  "decoder_start_token_id": null,48  "dtype": "float32",49  "encoder": {50    "_name_or_path": "/tmp/tmpk7yofmau/encoder/BertModel",51    "add_cross_attention": false,52    "architectures": [53      "BertModel"54    ],55    "attention_probs_dropout_prob": 0.1,56    "bos_token_id": null,57    "chunk_size_feed_forward": 0,58    "classifier_dropout": null,59    "dtype": "float32",60    "eos_token_id": null,61    "hidden_act": "gelu",62    "hidden_dropout_prob": 0.1,63    "hidden_size": 32,64    "id2label": {65      "0": "LABEL_0",66      "1": "LABEL_1"67    },68    "initializer_range": 0.02,69    "intermediate_size": 37,70    "is_decoder": false,71    "is_encoder_decoder": false,72    "label2id": {73      "LABEL_0": 0,74      "LABEL_1": 175    },76    "layer_norm_eps": 1e-12,77    "max_position_embeddings": 512,78    "model_type": "bert",79    "num_attention_heads": 4,80    "num_hidden_layers": 2,81    "output_attentions": false,82    "output_hidden_states": false,83    "pad_token_id": 0,84    "problem_type": null,85    "return_dict": true,86    "tie_word_embeddings": true,87    "type_vocab_size": 16,88    "use_cache": true,89    "vocab_size": 3052290  },91  "is_encoder_decoder": true,92  "model_type": "encoder-decoder",93  "pad_token_id": null,94  "transformers_version": "5.16.0.dev0"95}96