pirxus/wav2vec2_s2t_encoder

1{2  "activation_dropout": 0.1,3  "adapter_attn_dim": null,4  "adapter_kernel_size": 3,5  "adapter_stride": 2,6  "add_adapter": false,7  "apply_spec_augment": false,8  "architectures": [9    "Wav2Vec2ForCTC"10  ],11  "attention_dropout": 0.1,12  "bos_token_id": 1,13  "classifier_proj_size": 256,14  "codevector_dim": 256,15  "contrastive_logits_temperature": 0.1,16  "conv_bias": false,17  "conv_dim": [18    256,19    25620  ],21  "conv_kernel": [22    3,23    324  ],25  "conv_stride": [26    2,27    228  ],29  "ctc_loss_reduction": "sum",30  "ctc_zero_infinity": true,31  "diversity_loss_weight": 0.1,32  "do_stable_layer_norm": false,33  "eos_token_id": 2,34  "feat_extract_activation": "gelu",35  "feat_extract_norm": "group",36  "feat_proj_dropout": 0.0,37  "feat_quantizer_dropout": 0.0,38  "final_dropout": 0.1,39  "hidden_act": "gelu",40  "hidden_dropout": 0.1,41  "hidden_size": 256,42  "initializer_range": 0.02,43  "intermediate_size": 2048,44  "layer_norm_eps": 1e-05,45  "layerdrop": 0.1,46  "mask_feature_length": 10,47  "mask_feature_min_masks": 0,48  "mask_feature_prob": 0.0,49  "mask_time_length": 10,50  "mask_time_min_masks": 2,51  "mask_time_prob": 0.05,52  "model_type": "wav2vec2",53  "num_adapter_layers": 3,54  "num_attention_heads": 4,55  "num_codevector_groups": 2,56  "num_codevectors_per_group": 320,57  "num_conv_pos_embedding_groups": 16,58  "num_conv_pos_embeddings": 128,59  "num_feat_extract_layers": 2,60  "num_hidden_layers": 12,61  "num_negatives": 100,62  "output_hidden_size": 256,63  "pad_token_id": 0,64  "proj_codevector_dim": 256,65  "tdnn_dilation": [66    1,67    2,68    3,69    1,70    171  ],72  "tdnn_dim": [73    512,74    512,75    512,76    512,77    150078  ],79  "tdnn_kernel": [80    5,81    3,82    3,83    1,84    185  ],86  "torch_dtype": "float32",87  "transformers_version": "4.36.2",88  "use_weighted_layer_sum": false,89  "vocab_size": 32,90  "xvector_output_dim": 51291}92