{ "_name_or_path": "/content/DarijaBERT_mix/", "architectures": [ "BertForSequenceClassification" ], "attention_probs_dropout_prob": 0.1, "classifier_dropout": null, "cls_token": "[CLS]", "do_lower_case": true, "hidden_act": "gelu", "hidden_dropout_prob": 0.1, "hidden_size": 768, "initializer_range": 0.02, "intermediate_size": 3072, "layer_norm_eps": 1e-12, "max_len": 128, "max_position_embeddings": 512, "model_max_length": 128, "model_type": "bert", "num_attention_heads": 12, "num_hidden_layers": 12, "pad_token": "[PAD]", "pad_token_id": 0, "position_embedding_type": "absolute", "sep_token": "[SEP]", "torch_dtype": "float32", "transformers_version": "4.19.2", "type_vocab_size": 2, "unk_token": "[UNK]", "use_cache": true, "vocab_size": 160000 }