pruna-test/whisper-v3-tiny-random

1{2  "_name_or_path": "openai/whisper-large-v3",3  "activation_dropout": 0.0,4  "activation_function": "gelu",5  "apply_spec_augment": false,6  "architectures": [7    "WhisperForConditionalGeneration"8  ],9  "attention_dropout": 0.0,10  "begin_suppress_tokens": null,11  "bos_token_id": 50257,12  "classifier_proj_size": 256,13  "d_model": 8,14  "decoder_attention_heads": 2,15  "decoder_ffn_dim": 16,16  "decoder_layerdrop": 0.0,17  "decoder_layers": 2,18  "decoder_start_token_id": 50258,19  "dropout": 0.0,20  "encoder_attention_heads": 2,21  "encoder_ffn_dim": 16,22  "encoder_layerdrop": 0.0,23  "encoder_layers": 2,24  "eos_token_id": 50257,25  "init_std": 0.02,26  "is_encoder_decoder": true,27  "mask_feature_length": 10,28  "mask_feature_min_masks": 0,29  "mask_feature_prob": 0.0,30  "mask_time_length": 10,31  "mask_time_min_masks": 2,32  "mask_time_prob": 0.05,33  "max_length": null,34  "max_source_positions": 1500,35  "max_target_positions": 448,36  "median_filter_width": 7,37  "model_type": "whisper",38  "num_hidden_layers": 2,39  "num_mel_bins": 128,40  "pad_token_id": 50256,41  "scale_embedding": false,42  "torch_dtype": "float16",43  "transformers_version": "4.46.0",44  "use_cache": true,45  "use_weighted_layer_sum": false,46  "vocab_size": 5186647}48