hf-tiny-model-private/tiny-random-FlaubertForSequenceClassification

1{2  "_name_or_path": "tiny_models/flaubert/FlaubertForSequenceClassification",3  "architectures": [4    "FlaubertForSequenceClassification"5  ],6  "asm": false,7  "attention_dropout": 0.1,8  "bos_index": 0,9  "bos_token_id": 0,10  "causal": false,11  "dropout": 0.1,12  "emb_dim": 32,13  "embed_init_std": 0.02209708691207961,14  "end_n_top": 5,15  "eos_index": 1,16  "gelu_activation": true,17  "init_std": 0.02,18  "initializer_range": 0.02,19  "is_encoder": true,20  "lang_id": 0,21  "layer_norm_eps": 1e-12,22  "layerdrop": 0.0,23  "mask_index": 5,24  "mask_token_id": 5,25  "max_position_embeddings": 512,26  "model_type": "flaubert",27  "n_heads": 4,28  "n_langs": 2,29  "n_layers": 5,30  "n_special": 0,31  "pad_index": 2,32  "pad_token_id": 2,33  "pre_norm": false,34  "sinusoidal_embeddings": false,35  "start_n_top": 5,36  "summary_activation": null,37  "summary_first_dropout": 0.1,38  "summary_proj_to_labels": true,39  "summary_type": "last",40  "summary_use_proj": true,41  "torch_dtype": "float32",42  "transformers_version": "4.28.0.dev0",43  "unk_index": 3,44  "use_lang_emb": true,45  "use_proj": null,46  "vocab_size": 6754247}48