webis/sparse-cross-encoder-4-512

1{2  "_name_or_path": "cross-encoder/ms-marco-MiniLM-L-6-v2",3  "architectures": [4    "ListformerModelForSequenceClassification"5  ],6  "attention_probs_dropout_prob": 0.1,7  "attention_window_size": 4,8  "cls_doc_attention": true,9  "cls_query_attention": true,10  "cls_token_id": 101,11  "depth": 100,12  "doc_cls_attention": true,13  "doc_query_attention": true,14  "gradient_checkpointing": false,15  "hidden_act": "gelu",16  "hidden_dropout_prob": 0.1,17  "hidden_size": 384,18  "id2label": {19    "0": "LABEL_0"20  },21  "initializer_range": 0.02,22  "input_type_embedding": false,23  "intermediate_size": 1536,24  "label2id": {25    "LABEL_0": 026  },27  "layer_norm_eps": 1e-12,28  "max_position_embeddings": 512,29  "model_type": "bert",30  "normalize_output": false,31  "num_attention_heads": 12,32  "num_hidden_layers": 6,33  "pad_token_id": 0,34  "position_embedding_type": "absolute",35  "query_cls_attention": false,36  "query_doc_attention": false,37  "sbert_ce_default_activation_function": "torch.nn.modules.linear.Identity",38  "sep_token_id": 102,39  "token_type_embedding": true,40  "torch_dtype": "float32",41  "transformers_version": "4.28.1",42  "type_vocab_size": 2,43  "use_cache": true,44  "vocab_size": 3052245}46