LocalAI-io/whisper-base-it
014
1{2 "activation_dropout": 0.0,3 "activation_function": "gelu",4 "apply_spec_augment": false,5 "architectures": [6 "WhisperForConditionalGeneration"7 ],8 "attention_dropout": 0.0,9 "begin_suppress_tokens": null,10 "bos_token_id": 50257,11 "classifier_proj_size": 256,12 "d_model": 512,13 "decoder_attention_heads": 8,14 "decoder_ffn_dim": 2048,15 "decoder_layerdrop": 0.0,16 "decoder_layers": 6,17 "decoder_start_token_id": 50258,18 "dropout": 0.0,19 "dtype": "float32",20 "encoder_attention_heads": 8,21 "encoder_ffn_dim": 2048,22 "encoder_layerdrop": 0.0,23 "encoder_layers": 6,24 "eos_token_id": 50257,25 "forced_decoder_ids": [26 [27 1,28 5025929 ],30 [31 2,32 5035933 ],34 [35 3,36 5036337 ]38 ],39 "init_std": 0.02,40 "is_encoder_decoder": true,41 "mask_feature_length": 10,42 "mask_feature_min_masks": 0,43 "mask_feature_prob": 0.0,44 "mask_time_length": 10,45 "mask_time_min_masks": 2,46 "mask_time_prob": 0.05,47 "max_length": null,48 "max_source_positions": 1500,49 "max_target_positions": 448,50 "median_filter_width": 7,51 "model_type": "whisper",52 "num_hidden_layers": 6,53 "num_mel_bins": 80,54 "pad_token_id": 50257,55 "scale_embedding": false,56 "transformers_version": "4.57.6",57 "use_cache": true,58 "use_weighted_layer_sum": false,59 "vocab_size": 5186560}61 