├── .gitignore ├── requirements.txt ├── main.py ├── readme.md ├── download.py ├── download_wget.txt └── main.iipynb.ipynb /.gitignore: -------------------------------------------------------------------------------- 1 | .idea/ -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | wget -------------------------------------------------------------------------------- /main.py: -------------------------------------------------------------------------------- 1 | import nemo.collections.asr as nemo_asr 2 | 3 | 4 | model = nemo_asr.models.ASRModel.from_pretrained('QuartzNet15x5Base-En') 5 | 6 | 7 | -------------------------------------------------------------------------------- /readme.md: -------------------------------------------------------------------------------- 1 | This repository aims to download all nemo pretrained models. 2 | 3 | You can do this with wget package in python 4 | 5 | 6 | ### How to use 7 | 8 | #### Environment Setup 9 | > python 10 | > pip install wget 11 | > 12 | 13 | #### Download 14 | > python 15 | > python download.py 16 | > 17 | 18 | 19 | As you can see very simple. 20 | -------------------------------------------------------------------------------- /download.py: -------------------------------------------------------------------------------- 1 | import wget 2 | import os 3 | import logging 4 | from time import sleep 5 | 6 | 7 | def download(wget_uri, destination_file): 8 | i = 0 9 | max_attempts = 3 10 | while i < max_attempts: 11 | i += 1 12 | try: 13 | 14 | wget.download(wget_uri, str(destination_file)) 15 | if os.path.exists(destination_file): 16 | return destination_file 17 | else: 18 | return "" 19 | except: 20 | logging.info(f"Download from cloud failed. Attempt {i} of {max_attempts}") 21 | sleep(0.05) 22 | continue 23 | 24 | if __name__ == '__main__': 25 | lines = [] 26 | with open('download_wget.txt', 'r') as file: 27 | lines = file.readlines() 28 | for url in lines: 29 | print('Downloading: ', url) 30 | dest = 'models/' + url.split('/')[-1].replace('\n', '') 31 | download(url, dest) 32 | 33 | -------------------------------------------------------------------------------- /download_wget.txt: -------------------------------------------------------------------------------- 1 | https://api.ngc.nvidia.com/v2/models/nvidia/nemospeechmodels/versions/1.0.0a5/files/QuartzNet15x5Base-En.nemo 2 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/asr_talknet_aligner/versions/1.0.0rc1/files/qn5x5_libri_tts_phonemes.nemo 3 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x1x64_v1/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x1x64_v1.nemo 4 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x1x64_v2/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x1x64_v2.nemo 5 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x1x64_v2_subset_task/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x1x64_v2_subset_task.nemo 6 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x2x64_v1/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x2x64_v1.nemo 7 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x2x64_v2/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x2x64_v2.nemo 8 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x2x64_v2_subset_task/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x2x64_v2_subset_task.nemo 9 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_be_conformer_ctc_large/versions/1.12.0/files/stt_be_conformer_ctc_large.nemo 10 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_be_conformer_transducer_large/versions/1.12.0/files/stt_be_conformer_transducer_large.nemo 11 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_by_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_by_fastconformer_hybrid_large_pc.nemo 12 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ca_conformer_ctc_large/versions/1.11.0/files/stt_ca_conformer_ctc_large.nemo 13 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ca_conformer_transducer_large/versions/1.11.0/files/stt_ca_conformer_transducer_large.nemo 14 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ca_quartznet15x5/versions/1.0.0rc1/files/stt_ca_quartznet15x5.nemo 15 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_citrinet_1024/versions/1.5.0/files/stt_de_citrinet_1024.nemo 16 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_conformer_ctc_large/versions/1.5.0/files/stt_de_conformer_ctc_large.nemo 17 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_conformer_transducer_large/versions/1.5.0/files/stt_de_conformer_transducer_large.nemo 18 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_contextnet_1024/versions/1.4.0/files/stt_de_contextnet_1024.nemo 19 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_de_fastconformer_hybrid_large_pc.nemo 20 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_quartznet15x5/versions/1.0.0rc1/files/stt_de_quartznet15x5.nemo 21 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_1024/versions/1.0.0rc1/files/stt_en_citrinet_1024.nemo 22 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_1024_gamma_0_25/versions/1.0.0/files/stt_en_citrinet_1024_gamma_0_25.nemo 23 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_256/versions/1.0.0rc1/files/stt_en_citrinet_256.nemo 24 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_256_gamma_0_25/versions/1.0.0/files/stt_en_citrinet_256_gamma_0_25.nemo 25 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_512/versions/1.0.0rc1/files/stt_en_citrinet_512.nemo 26 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_512_gamma_0_25/versions/1.0.0/files/stt_en_citrinet_512_gamma_0_25.nemo 27 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_large/versions/1.10.0/files/stt_en_conformer_ctc_large.nemo 28 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_large_ls/versions/1.0.0/files/stt_en_conformer_ctc_large_ls.nemo 29 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_medium/versions/1.6.0/files/stt_en_conformer_ctc_medium.nemo 30 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_medium_ls/versions/1.0.0/files/stt_en_conformer_ctc_medium_ls.nemo 31 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_small/versions/1.6.0/files/stt_en_conformer_ctc_small.nemo 32 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_small_ls/versions/1.0.0/files/stt_en_conformer_ctc_small_ls.nemo 33 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_xlarge/versions/1.10.0/files/stt_en_conformer_ctc_xlarge.nemo 34 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_large/versions/1.10.0/files/stt_en_conformer_transducer_large.nemo 35 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_large_ls/versions/1.8.0/files/stt_en_conformer_transducer_large_ls.nemo 36 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_medium/versions/1.6.0/files/stt_en_conformer_transducer_medium.nemo 37 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_small/versions/1.6.0/files/stt_en_conformer_transducer_small.nemo 38 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_xlarge/versions/1.10.0/files/stt_en_conformer_transducer_xlarge.nemo 39 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_xxlarge/versions/1.8.0/files/stt_en_conformer_transducer_xxlarge.nemo 40 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_1024/versions/1.9.0/files/stt_en_contextnet_1024.nemo 41 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_1024_mls/versions/1.0.0/files/stt_en_contextnet_1024_mls.nemo 42 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_256/versions/1.6.0/files/stt_en_contextnet_256.nemo 43 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_256_mls/versions/1.0.0/files/stt_en_contextnet_256_mls.nemo 44 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_512/versions/1.6.0/files/stt_en_contextnet_512.nemo 45 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_512_mls/versions/1.0.0/files/stt_en_contextnet_512_mls.nemo 46 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_ctc_large/versions/1.0.0/files/stt_en_fastconformer_ctc_large.nemo 47 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_ctc_xlarge/versions/1.20.0/files/stt_en_fastconformer_ctc_xlarge.nemo 48 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_ctc_xxlarge/versions/1.20.1/files/stt_en_fastconformer_ctc_xxlarge.nemo 49 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_en_fastconformer_hybrid_large_pc.nemo 50 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_1040ms/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_1040ms.nemo 51 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_480ms/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_480ms.nemo 52 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_80ms/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_80ms.nemo 53 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_multi/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_multi.nemo 54 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_transducer_large/versions/1.0.0/files/stt_en_fastconformer_transducer_large.nemo 55 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_transducer_xlarge/versions/1.20.1/files/stt_en_fastconformer_transducer_xlarge.nemo 56 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_transducer_xxlarge/versions/1.20.1/files/stt_en_fastconformer_transducer_xxlarge.nemo 57 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_jasper10x5dr/versions/1.0.0rc1/files/stt_en_jasper10x5dr.nemo 58 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_quartznet15x5/versions/1.0.0rc1/files/stt_en_quartznet15x5.nemo 59 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_large_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_large_ls.nemo 60 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_medium_large_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_medium_large_ls.nemo 61 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_medium_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_medium_ls.nemo 62 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_small_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_small_ls.nemo 63 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_small_medium_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_small_medium_ls.nemo 64 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_xsmall_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_xsmall_ls.nemo 65 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_ctc_large/versions/1.0.0/files/stt_enes_conformer_ctc_large.nemo 66 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_ctc_large_codesw/versions/1.0.0/files/stt_enes_conformer_ctc_large_codesw.nemo 67 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_transducer_large/versions/1.0.0/files/stt_enes_conformer_transducer_large.nemo 68 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_transducer_large_codesw/versions/1.0.0/files/stt_enes_conformer_transducer_large_codesw.nemo 69 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_contextnet_large/versions/1.0.0/files/stt_enes_contextnet_large.nemo 70 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_eo_conformer_ctc_large/versions/1.14.0/files/stt_eo_conformer_ctc_large.nemo 71 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_eo_conformer_transducer_large/versions/1.14.0/files/stt_eo_conformer_transducer_large.nemo 72 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_citrinet_1024_gamma_0_25/versions/1.8.0/files/stt_es_citrinet_1024_gamma_0_25.nemo 73 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_citrinet_512/versions/1.0.0/files/stt_es_citrinet_512.nemo 74 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_conformer_ctc_large/versions/1.8.0/files/stt_es_conformer_ctc_large.nemo 75 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_conformer_transducer_large/versions/1.8.0/files/stt_es_conformer_transducer_large.nemo 76 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_contextnet_1024/versions/1.8.0/files/stt_es_contextnet_1024.nemo 77 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_es_fastconformer_hybrid_large_pc.nemo 78 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_quartznet15x5/versions/1.0.0rc1/files/stt_es_quartznet15x5.nemo 79 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_citrinet_1024_gamma_0_25/versions/1.5/files/stt_fr_citrinet_1024_gamma_0_25.nemo 80 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_conformer_ctc_large/versions/1.5.1/files/stt_fr_conformer_ctc_large.nemo 81 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_conformer_transducer_large/versions/1.5/files/stt_fr_conformer_transducer_large.nemo 82 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_contextnet_1024/versions/1.5/files/stt_fr_contextnet_1024.nemo 83 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_fr_fastconformer_hybrid_large_pc.nemo 84 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_citrinet_1024_gamma_0_25/versions/1.5/files/stt_fr_no_hyphen_citrinet_1024_gamma_0_25.nemo 85 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_conformer_ctc_large/versions/1.5.1/files/stt_fr_no_hyphen_conformer_ctc_large.nemo 86 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_quartznet15x5/versions/1.0.0rc1/files/stt_fr_quartznet15x5.nemo 87 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hi_conformer_ctc_medium/versions/1.6.0/files/stt_hi_conformer_ctc_medium.nemo 88 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hr_conformer_ctc_large/versions/1.11.0/files/stt_hr_conformer_ctc_large.nemo 89 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hr_conformer_transducer_large/versions/1.11.0/files/stt_hr_conformer_transducer_large.nemo 90 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hr_fastconformer_hybrid_large_pc/versions/1.21.0/files/FastConformer-Hybrid-Transducer-CTC-BPE-v256-averaged.nemo 91 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_conformer_ctc_large/versions/1.13.0/files/stt_it_conformer_ctc_large.nemo 92 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_conformer_transducer_large/versions/1.13.0/files/stt_it_conformer_transducer_large.nemo 93 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_fastconformer_hybrid_large_pc/versions/1.20.0/files/stt_it_fastconformer_hybrid_large_pc.nemo 94 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_quartznet15x5/versions/1.0.0rc1/files/stt_it_quartznet15x5.nemo 95 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_kab_conformer_transducer_large/versions/1.12.0/files/stt_kab_conformer_transducer_large.nemo 96 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_mr_conformer_ctc_medium/versions/1.6.0/files/stt_mr_conformer_ctc_medium.nemo 97 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_multilingual_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_multilingual_fastconformer_hybrid_large_pc.nemo 98 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_multilingual_fastconformer_hybrid_large_pc_blend_eu/versions/1.21.0/files/stt_multilingual_fastconformer_hybrid_large_pc_blend_eu.nemo 99 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_pl_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_pl_fastconformer_hybrid_large_pc.nemo 100 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_pl_quartznet15x5/versions/1.0.0rc1/files/stt_pl_quartznet15x5.nemo 101 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_conformer_ctc_large/versions/1.13.0/files/stt_ru_conformer_ctc_large.nemo 102 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_conformer_transducer_large/versions/1.13.0/files/stt_ru_conformer_transducer_large.nemo 103 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_ru_fastconformer_hybrid_large_pc.nemo 104 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_quartznet15x5/versions/1.0.0rc1/files/stt_ru_quartznet15x5.nemo 105 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_rw_conformer_ctc_large/versions/1.11.0/files/stt_rw_conformer_ctc_large.nemo 106 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_rw_conformer_transducer_large/versions/1.11.0/files/stt_rw_conformer_transducer_large.nemo 107 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ua_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_ua_fastconformer_hybrid_large_pc.nemo 108 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_citrinet_1024_gamma_0_25/versions/1.0.0/files/stt_zh_citrinet_1024_gamma_0_25.nemo 109 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_citrinet_512/versions/1.0.0rc1/files/stt_zh_citrinet_512.nemo 110 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_conformer_transducer_large/versions/1.8.0/files/stt_zh_conformer_transducer_large.nemo 111 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_marblenet/versions/1.0.0rc1/files/vad_marblenet.nemo 112 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_multilingual_frame_marblenet/versions/1.20.0/files/vad_multilingual_frame_marblenet.nemo 113 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_multilingual_marblenet/versions/1.10.0/files/vad_multilingual_marblenet.nemo 114 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_telephony_marblenet/versions/1.0.0rc1/files/vad_telephony_marblenet.nemo 115 | https://api.ngc.nvidia.com/v2/models/nvidia/nemospeechmodels/versions/1.0.0a5/files/QuartzNet15x5Base-En.nemo 116 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_quartznet15x5/versions/1.0.0rc1/files/stt_en_quartznet15x5.nemo 117 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_jasper10x5dr/versions/1.0.0rc1/files/stt_en_jasper10x5dr.nemo 118 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ca_quartznet15x5/versions/1.0.0rc1/files/stt_ca_quartznet15x5.nemo 119 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_quartznet15x5/versions/1.0.0rc1/files/stt_it_quartznet15x5.nemo 120 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_quartznet15x5/versions/1.0.0rc1/files/stt_fr_quartznet15x5.nemo 121 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_quartznet15x5/versions/1.0.0rc1/files/stt_es_quartznet15x5.nemo 122 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_quartznet15x5/versions/1.0.0rc1/files/stt_de_quartznet15x5.nemo 123 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_pl_quartznet15x5/versions/1.0.0rc1/files/stt_pl_quartznet15x5.nemo 124 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_quartznet15x5/versions/1.0.0rc1/files/stt_ru_quartznet15x5.nemo 125 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_citrinet_512/versions/1.0.0rc1/files/stt_zh_citrinet_512.nemo 126 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_citrinet_1024_gamma_0_25/versions/1.0.0/files/stt_zh_citrinet_1024_gamma_0_25.nemo 127 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_citrinet_1024_gamma_0_25/versions/1.0.0/files/stt_zh_citrinet_1024_gamma_0_25.nemo 128 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/asr_talknet_aligner/versions/1.0.0rc1/files/qn5x5_libri_tts_phonemes.nemo 129 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_256/versions/1.0.0rc1/files/stt_en_citrinet_256.nemo 130 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_512/versions/1.0.0rc1/files/stt_en_citrinet_512.nemo 131 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_1024/versions/1.0.0rc1/files/stt_en_citrinet_1024.nemo 132 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_256_gamma_0_25/versions/1.0.0/files/stt_en_citrinet_256_gamma_0_25.nemo 133 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_512_gamma_0_25/versions/1.0.0/files/stt_en_citrinet_512_gamma_0_25.nemo 134 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_citrinet_1024_gamma_0_25/versions/1.0.0/files/stt_en_citrinet_1024_gamma_0_25.nemo 135 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_citrinet_512/versions/1.0.0/files/stt_es_citrinet_512.nemo 136 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_citrinet_1024/versions/1.5.0/files/stt_de_citrinet_1024.nemo 137 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_citrinet_1024_gamma_0_25/versions/1.5/files/stt_fr_citrinet_1024_gamma_0_25.nemo 138 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_citrinet_1024_gamma_0_25/versions/1.5/files/stt_fr_no_hyphen_citrinet_1024_gamma_0_25.nemo 139 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_citrinet_1024_gamma_0_25/versions/1.8.0/files/stt_es_citrinet_1024_gamma_0_25.nemo 140 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_small/versions/1.6.0/files/stt_en_conformer_ctc_small.nemo 141 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_medium/versions/1.6.0/files/stt_en_conformer_ctc_medium.nemo 142 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_large/versions/1.10.0/files/stt_en_conformer_ctc_large.nemo 143 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_xlarge/versions/1.10.0/files/stt_en_conformer_ctc_xlarge.nemo 144 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_xsmall_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_xsmall_ls.nemo 145 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_small_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_small_ls.nemo 146 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_small_medium_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_small_medium_ls.nemo 147 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_medium_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_medium_ls.nemo 148 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_medium_large_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_medium_large_ls.nemo 149 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_squeezeformer_ctc_large_ls/versions/1.13.0/files/stt_en_squeezeformer_ctc_large_ls.nemo 150 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_small_ls/versions/1.0.0/files/stt_en_conformer_ctc_small_ls.nemo 151 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_medium_ls/versions/1.0.0/files/stt_en_conformer_ctc_medium_ls.nemo 152 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_ctc_large_ls/versions/1.0.0/files/stt_en_conformer_ctc_large_ls.nemo 153 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_conformer_ctc_large/versions/1.5.1/files/stt_fr_conformer_ctc_large.nemo 154 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_conformer_ctc_large/versions/1.5.1/files/stt_fr_no_hyphen_conformer_ctc_large.nemo 155 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_conformer_ctc_large/versions/1.5.0/files/stt_de_conformer_ctc_large.nemo 156 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_conformer_ctc_large/versions/1.8.0/files/stt_es_conformer_ctc_large.nemo 157 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hi_conformer_ctc_medium/versions/1.6.0/files/stt_hi_conformer_ctc_medium.nemo 158 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_mr_conformer_ctc_medium/versions/1.6.0/files/stt_mr_conformer_ctc_medium.nemo 159 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_ctc_large/versions/1.0.0/files/stt_enes_conformer_ctc_large.nemo 160 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ca_conformer_ctc_large/versions/1.11.0/files/stt_ca_conformer_ctc_large.nemo 161 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_rw_conformer_ctc_large/versions/1.11.0/files/stt_rw_conformer_ctc_large.nemo 162 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_ctc_large_codesw/versions/1.0.0/files/stt_enes_conformer_ctc_large_codesw.nemo 163 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_be_conformer_ctc_large/versions/1.12.0/files/stt_be_conformer_ctc_large.nemo 164 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hr_conformer_ctc_large/versions/1.11.0/files/stt_hr_conformer_ctc_large.nemo 165 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_conformer_ctc_large/versions/1.13.0/files/stt_it_conformer_ctc_large.nemo 166 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_conformer_ctc_large/versions/1.13.0/files/stt_ru_conformer_ctc_large.nemo 167 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_eo_conformer_ctc_large/versions/1.14.0/files/stt_eo_conformer_ctc_large.nemo 168 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_ctc_large/versions/1.0.0/files/stt_en_fastconformer_ctc_large.nemo 169 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_ctc_xlarge/versions/1.20.0/files/stt_en_fastconformer_ctc_xlarge.nemo 170 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_ctc_xxlarge/versions/1.20.1/files/stt_en_fastconformer_ctc_xxlarge.nemo 171 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_multilingual_marblenet/versions/1.10.0/files/vad_multilingual_marblenet.nemo 172 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_telephony_marblenet/versions/1.0.0rc1/files/vad_telephony_marblenet.nemo 173 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_marblenet/versions/1.0.0rc1/files/vad_marblenet.nemo 174 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x1x64_v1/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x1x64_v1.nemo 175 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x2x64_v1/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x2x64_v1.nemo 176 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x1x64_v2/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x1x64_v2.nemo 177 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x2x64_v2/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x2x64_v2.nemo 178 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x1x64_v2_subset_task/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x1x64_v2_subset_task.nemo 179 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/commandrecognition_en_matchboxnet3x2x64_v2_subset_task/versions/1.0.0rc1/files/commandrecognition_en_matchboxnet3x2x64_v2_subset_task.nemo 180 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/diar_msdd_telephonic/versions/1.0.1/files/diar_msdd_telephonic.nemo 181 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/vad_multilingual_frame_marblenet/versions/1.20.0/files/vad_multilingual_frame_marblenet.nemo 182 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_en_fastconformer_hybrid_large_pc.nemo 183 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_de_fastconformer_hybrid_large_pc.nemo 184 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_fastconformer_hybrid_large_pc/versions/1.20.0/files/stt_it_fastconformer_hybrid_large_pc.nemo 185 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_es_fastconformer_hybrid_large_pc.nemo 186 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hr_fastconformer_hybrid_large_pc/versions/1.21.0/files/FastConformer-Hybrid-Transducer-CTC-BPE-v256-averaged.nemo 187 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ua_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_ua_fastconformer_hybrid_large_pc.nemo 188 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_pl_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_pl_fastconformer_hybrid_large_pc.nemo 189 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_by_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_by_fastconformer_hybrid_large_pc.nemo 190 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_ru_fastconformer_hybrid_large_pc.nemo 191 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_fr_fastconformer_hybrid_large_pc.nemo 192 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_multilingual_fastconformer_hybrid_large_pc_blend_eu/versions/1.21.0/files/stt_multilingual_fastconformer_hybrid_large_pc_blend_eu.nemo 193 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_multilingual_fastconformer_hybrid_large_pc/versions/1.21.0/files/stt_multilingual_fastconformer_hybrid_large_pc.nemo 194 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_80ms/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_80ms.nemo 195 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_480ms/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_480ms.nemo 196 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_1040ms/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_1040ms.nemo 197 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_hybrid_large_streaming_multi/versions/1.20.0/files/stt_en_fastconformer_hybrid_large_streaming_multi.nemo 198 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_256/versions/1.6.0/files/stt_en_contextnet_256.nemo 199 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_512/versions/1.6.0/files/stt_en_contextnet_512.nemo 200 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_1024/versions/1.9.0/files/stt_en_contextnet_1024.nemo 201 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_256_mls/versions/1.0.0/files/stt_en_contextnet_256_mls.nemo 202 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_512_mls/versions/1.0.0/files/stt_en_contextnet_512_mls.nemo 203 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_contextnet_1024_mls/versions/1.0.0/files/stt_en_contextnet_1024_mls.nemo 204 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_small/versions/1.6.0/files/stt_en_conformer_transducer_small.nemo 205 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_medium/versions/1.6.0/files/stt_en_conformer_transducer_medium.nemo 206 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_large/versions/1.10.0/files/stt_en_conformer_transducer_large.nemo 207 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_large_ls/versions/1.8.0/files/stt_en_conformer_transducer_large_ls.nemo 208 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_xlarge/versions/1.10.0/files/stt_en_conformer_transducer_xlarge.nemo 209 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_conformer_transducer_xxlarge/versions/1.8.0/files/stt_en_conformer_transducer_xxlarge.nemo 210 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_contextnet_1024/versions/1.4.0/files/stt_de_contextnet_1024.nemo 211 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_contextnet_1024/versions/1.5/files/stt_fr_contextnet_1024.nemo 212 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_contextnet_1024/versions/1.8.0/files/stt_es_contextnet_1024.nemo 213 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_de_conformer_transducer_large/versions/1.5.0/files/stt_de_conformer_transducer_large.nemo 214 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_fr_conformer_transducer_large/versions/1.5/files/stt_fr_conformer_transducer_large.nemo 215 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_es_conformer_transducer_large/versions/1.8.0/files/stt_es_conformer_transducer_large.nemo 216 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_transducer_large/versions/1.0.0/files/stt_enes_conformer_transducer_large.nemo 217 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_contextnet_large/versions/1.0.0/files/stt_enes_contextnet_large.nemo 218 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ca_conformer_transducer_large/versions/1.11.0/files/stt_ca_conformer_transducer_large.nemo 219 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_rw_conformer_transducer_large/versions/1.11.0/files/stt_rw_conformer_transducer_large.nemo 220 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_enes_conformer_transducer_large_codesw/versions/1.0.0/files/stt_enes_conformer_transducer_large_codesw.nemo 221 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_kab_conformer_transducer_large/versions/1.12.0/files/stt_kab_conformer_transducer_large.nemo 222 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_be_conformer_transducer_large/versions/1.12.0/files/stt_be_conformer_transducer_large.nemo 223 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_hr_conformer_transducer_large/versions/1.11.0/files/stt_hr_conformer_transducer_large.nemo 224 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_it_conformer_transducer_large/versions/1.13.0/files/stt_it_conformer_transducer_large.nemo 225 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_ru_conformer_transducer_large/versions/1.13.0/files/stt_ru_conformer_transducer_large.nemo 226 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_eo_conformer_transducer_large/versions/1.14.0/files/stt_eo_conformer_transducer_large.nemo 227 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_transducer_large/versions/1.0.0/files/stt_en_fastconformer_transducer_large.nemo 228 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_transducer_xlarge/versions/1.20.1/files/stt_en_fastconformer_transducer_xlarge.nemo 229 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_en_fastconformer_transducer_xxlarge/versions/1.20.1/files/stt_en_fastconformer_transducer_xxlarge.nemo 230 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/stt_zh_conformer_transducer_large/versions/1.8.0/files/stt_zh_conformer_transducer_large.nemo 231 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/speakerverification_speakernet/versions/1.16.0/files/speakerverification_speakernet.nemo 232 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ecapa_tdnn/versions/1.16.0/files/ecapa_tdnn.nemo 233 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/titanet_large/versions/v1/files/titanet-l.nemo 234 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/langid_ambernet/versions/1.12.0/files/ambernet.nemo 235 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/titanet_small/versions/1.19.0/files/titanet-s.nemo 236 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/diar_msdd_telephonic/versions/1.0.1/files/diar_msdd_telephonic.nemo 237 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ssl_en_conformer_large/versions/1.10.1/files/ssl_en_conformer_large.nemo 238 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ssl_en_conformer_xlarge/versions/1.10.0/files/ssl_en_conformer_xlarge.nemo 239 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/bertbaseuncased/versions/1.0.0rc1/files/bertbaseuncased.nemo 240 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/bertlargeuncased/versions/1.0.0rc1/files/bertlargeuncased.nemo 241 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/neural_text_normalization_t5/versions/1.5.0/files/neural_text_normalization_t5_decoder.nemo 242 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_en_t5/versions/1.11.0/files/itn_en_t5_decoder.nemo 243 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/neural_text_normalization_t5/versions/1.5.0/files/neural_text_normalization_t5_tagger.nemo 244 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_en_t5/versions/1.11.0/files/itn_en_t5_tagger.nemo 245 | https://api.ngc.nvidia.com/v2/models/nvidia/nemonlpmodels/versions/1.0.0a5/files/Joint_Intent_Slot_Assistant.nemo 246 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_de_transformer12x2/versions/1.0.0rc1/files/nmt_en_de_transformer12x2.nemo 247 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_de_en_transformer12x2/versions/1.0.0rc1/files/nmt_de_en_transformer12x2.nemo 248 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_es_transformer12x2/versions/1.0.0rc1/files/nmt_en_es_transformer12x2.nemo 249 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_es_en_transformer12x2/versions/1.0.0rc1/files/nmt_es_en_transformer12x2.nemo 250 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_fr_transformer12x2/versions/1.0.0rc1/files/nmt_en_fr_transformer12x2.nemo 251 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_fr_en_transformer12x2/versions/1.0.0rc1/files/nmt_fr_en_transformer12x2.nemo 252 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_ru_transformer6x6/versions/1.0.0rc1/files/nmt_en_ru_transformer6x6.nemo 253 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_ru_en_transformer6x6/versions/1.0.0rc1/files/nmt_ru_en_transformer6x6.nemo 254 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_zh_en_transformer6x6/versions/1.0.0rc1/files/nmt_zh_en_transformer6x6.nemo 255 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_zh_transformer6x6/versions/1.0.0rc1/files/nmt_en_zh_transformer6x6.nemo 256 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_hi_en_transformer12x2/versions/v1.0.0/files/nmt_hi_en_transformer12x2.nemo 257 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_hi_transformer12x2/versions/v1.0.0/files/nmt_en_hi_transformer12x2.nemo 258 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_deesfr_en_transformer12x2/versions/1.2.0/files/mnmt_deesfr_en_transformer12x2.nemo 259 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_deesfr_en_transformer24x6/versions/1.2.0/files/mnmt_deesfr_en_transformer24x6.nemo 260 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_deesfr_en_transformer6x6/versions/1.2.0/files/mnmt_deesfr_en_transformer6x6.nemo 261 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformer12x2/versions/1.2.0/files/mnmt_en_deesfr_transformer12x2.nemo 262 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformer24x6/versions/1.2.0/files/mnmt_en_deesfr_transformer24x6.nemo 263 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformer6x6/versions/1.2.0/files/mnmt_en_deesfr_transformer6x6.nemo 264 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformerbase/versions/1.2.0/files/mnmt_en_deesfr_transformerbase.nemo 265 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_de_transformer24x6/versions/1.5/files/en_de_24x6.nemo 266 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_de_en_transformer24x6/versions/1.5/files/de_en_24x6.nemo 267 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_es_transformer24x6/versions/1.5/files/en_es_24x6.nemo 268 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_es_en_transformer24x6/versions/1.5/files/es_en_24x6.nemo 269 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_fr_transformer24x6/versions/1.5/files/en_fr_24x6.nemo 270 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_fr_en_transformer24x6/versions/1.5/files/fr_en_24x6.nemo 271 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_ru_transformer24x6/versions/1.5/files/en_ru_24x6.nemo 272 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_ru_en_transformer24x6/versions/1.5/files/ru_en_24x6.nemo 273 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_zh_transformer24x6/versions/1.5/files/en_zh_24x6.nemo 274 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_zh_en_transformer24x6/versions/1.5/files/zh_en_24x6.nemo 275 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/punctuation_en_bert/versions/1.0.0rc1/files/punctuation_en_bert.nemo 276 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/punctuation_en_distilbert/versions/1.0.0rc1/files/punctuation_en_distilbert.nemo 277 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_bertbase/versions/1.0.0rc1/files/qa_squadv1.1_bertbase.nemo 278 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_bertbase/versions/1.0.0rc1/files/qa_squadv2.0_bertbase.nemo 279 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_bertlarge/versions/1.0.0rc1/files/qa_squadv1.1_bertlarge.nemo 280 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_bertlarge/versions/1.0.0rc1/files/qa_squadv2.0_bertlarge.nemo 281 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_megatron_cased/versions/1.0.0rc1/files/qa_squadv1.1_megatron_cased.nemo 282 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_megatron_cased/versions/1.0.0rc1/files/qa_squadv2.0_megatron_cased.nemo 283 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_megatron_uncased/versions/1.0.0rc1/files/qa_squadv1.1_megatron_uncased.nemo 284 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_megatron_uncased/versions/1.0.0rc1/files/qa_squadv2.0_megatron_uncased.nemo 285 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_en_thutmose_bert/versions/1.9.0/files/itn_en_thutmose_bert.nemo 286 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_ru_thutmose_bert/versions/1.11.0/files/itn_ru_thutmose_bert.nemo 287 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ner_en_bert/versions/1.10/files/ner_en_bert.nemo 288 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/zeroshotintent_en_bert_base_uncased/versions/1.4.1/files/zeroshotintent_en_bert_base_uncased.nemo 289 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/zeroshotintent_en_megatron_uncased/versions/1.4.1/files/zeroshotintent_en_megatron_uncased.nemo 290 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_radtts_aligner/versions/ARPABET_1.11.0/files/Aligner.nemo 291 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_radtts_aligner/versions/IPA_1.13.0/files/Aligner.nemo 292 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_fastpitch/versions/1.8.1/files/tts_en_fastpitch_align.nemo 293 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_fastpitch/versions/IPA_1.13.0/files/tts_en_fastpitch_align_ipa.nemo 294 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_multispeaker_fastpitchhifigan/versions/1.10.0/files/tts_en_fastpitch_multispeaker.nemo 295 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_de_fastpitchhifigan/versions/1.15.0/files/tts_de_fastpitch_thorstens2102.nemo 296 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_de_fastpitchhifigan/versions/1.15.0/files/tts_de_fastpitch_thorstens2210.nemo 297 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_de_fastpitch_multispeaker_5/versions/1.11.0/files/tts_de_fastpitch_multispeaker_5.nemo 298 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_es_multispeaker_fastpitchhifigan/versions/1.15.0/files/tts_es_fastpitch_multispeaker.nemo 299 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_zh_fastpitch_hifigan_sfspeech/versions/1.15.0/files/tts_zh_fastpitch_sfspeech.nemo 300 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_fastpitch_spectrogram_enhancer_for_asr_finetuning/versions/1.20.0/files/tts_en_fastpitch_for_asr_finetuning.nemo 301 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_hifigan/versions/1.0.0rc1/files/tts_hifigan.nemo 302 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_lj_hifigan/versions/1.6.0/files/tts_en_lj_hifigan_ft_mixertts.nemo 303 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_lj_hifigan/versions/1.6.0/files/tts_en_lj_hifigan_ft_mixerttsx.nemo 304 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_multispeaker_fastpitchhifigan/versions/1.10.0/files/tts_en_hifitts_hifigan_ft_fastpitch.nemo 305 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_de_fastpitchhifigan/versions/1.15.0/files/tts_de_hifigan_thorstens2102.nemo 306 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_de_fastpitchhifigan/versions/1.15.0/files/tts_de_hifigan_thorstens2210.nemo 307 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_de_fastpitch_multispeaker_5/versions/1.11.0/files/tts_de_hui_hifigan_ft_fastpitch_multispeaker_5.nemo 308 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_es_multispeaker_fastpitchhifigan/versions/1.15.0/files/tts_es_hifigan_ft_fastpitch_multispeaker.nemo 309 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_zh_fastpitch_hifigan_sfspeech/versions/1.15.0/files/tts_zh_hifigan_sfspeech.nemo 310 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_lj_mixertts/versions/1.6.0/files/tts_en_lj_mixertts.nemo 311 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_lj_mixerttsx/versions/1.6.0/files/tts_en_lj_mixerttsx.nemo 312 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ssl_en_conformer_large/versions/1.10.1/files/ssl_en_conformer_large.nemo 313 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ssl_en_conformer_xlarge/versions/1.10.0/files/ssl_en_conformer_xlarge.nemo 314 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_fastpitch_spectrogram_enhancer_for_asr_finetuning/versions/1.20.0/files/tts_en_spectrogram_enhancer_for_asr_finetuning.nemo 315 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_tacotron2/versions/1.10.0/files/tts_en_tacotron2.nemo 316 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_lj_univnet/versions/1.7.0/files/tts_en_lj_univnet.nemo 317 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_libritts_univnet/versions/1.7.0/files/tts_en_libritts_multispeaker_univnet.nemo 318 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_lj_vits/versions/1.13.0/files/vits_ljspeech_fp16_full.nemo 319 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_en_hifitts_vits/versions/r1.15.0/files/vits_en_hifitts.nemo 320 | https://api.ngc.nvidia.com/v2/models/nvidia/nemo/tts_waveglow_88m/versions/1.0.0/files/tts_waveglow.nemo 321 | -------------------------------------------------------------------------------- /main.iipynb.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 1, 6 | "id": "initial_id", 7 | "metadata": { 8 | "collapsed": true, 9 | "ExecuteTime": { 10 | "end_time": "2024-10-31T22:32:15.081605900Z", 11 | "start_time": "2024-10-31T22:31:57.837995400Z" 12 | } 13 | }, 14 | "outputs": [ 15 | { 16 | "name": "stderr", 17 | "output_type": "stream", 18 | "text": [ 19 | "[NeMo W 2024-10-31 22:32:14 nemo_logging:349] C:\\Users\\Administrator\\anaconda3\\envs\\nemo\\lib\\site-packages\\nemo\\collections\\tts\\modules\\common.py:206: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.\n", 20 | " @amp.autocast(False)\n", 21 | " \n" 22 | ] 23 | } 24 | ], 25 | "source": [ 26 | "import nemo.collections.asr as nemo_asr\n", 27 | "import nemo.collections.nlp as nemo_nlp\n", 28 | "import nemo.collections.tts as nemo_tts" 29 | ] 30 | }, 31 | { 32 | "cell_type": "code", 33 | "execution_count": 50, 34 | "outputs": [], 35 | "source": [ 36 | "nlp_base_models = dir(nemo_nlp.models) " 37 | ], 38 | "metadata": { 39 | "collapsed": false, 40 | "ExecuteTime": { 41 | "end_time": "2024-11-01T04:26:43.853918200Z", 42 | "start_time": "2024-11-01T04:26:43.835124Z" 43 | } 44 | }, 45 | "id": "337b4b97e2a2fec3" 46 | }, 47 | { 48 | "cell_type": "code", 49 | "execution_count": 51, 50 | "outputs": [ 51 | { 52 | "data": { 53 | "text/plain": "['BERTLMModel',\n 'BertDPRModel',\n 'BertJointIRModel',\n 'DuplexDecoderModel',\n 'DuplexTaggerModel',\n 'DuplexTextNormalizationModel',\n 'EntityLinkingModel',\n 'GLUEModel',\n 'IntentSlotClassificationModel',\n 'MTEncDecModel',\n 'MegatronGPTPromptLearningModel',\n 'MultiLabelIntentSlotClassificationModel',\n 'PunctuationCapitalizationLexicalAudioModel',\n 'PunctuationCapitalizationModel',\n 'QAModel',\n 'SpellcheckingAsrCustomizationModel',\n 'Text2SparqlModel',\n 'TextClassificationModel',\n 'ThutmoseTaggerModel',\n 'TokenClassificationModel',\n 'TransformerLMModel',\n 'ZeroShotIntentModel',\n '__builtins__',\n '__cached__',\n '__doc__',\n '__file__',\n '__loader__',\n '__name__',\n '__package__',\n '__path__',\n '__spec__',\n 'duplex_text_normalization',\n 'enc_dec_nlp_model',\n 'entity_linking',\n 'glue_benchmark',\n 'information_retrieval',\n 'intent_slot_classification',\n 'language_modeling',\n 'machine_translation',\n 'nlp_model',\n 'question_answering',\n 'spellchecking_asr_customization',\n 'text2sparql',\n 'text_classification',\n 'text_normalization_as_tagging',\n 'token_classification',\n 'zero_shot_intent_recognition']" 54 | }, 55 | "execution_count": 51, 56 | "metadata": {}, 57 | "output_type": "execute_result" 58 | } 59 | ], 60 | "source": [ 61 | "nlp_base_models" 62 | ], 63 | "metadata": { 64 | "collapsed": false, 65 | "ExecuteTime": { 66 | "end_time": "2024-11-01T04:26:44.744205300Z", 67 | "start_time": "2024-11-01T04:26:44.728578100Z" 68 | } 69 | }, 70 | "id": "3ee7f56c3345d1bd" 71 | }, 72 | { 73 | "cell_type": "code", 74 | "execution_count": 52, 75 | "outputs": [ 76 | { 77 | "name": "stdout", 78 | "output_type": "stream", 79 | "text": [ 80 | "BERTLMModel\n", 81 | "bertbaseuncased\n", 82 | "bertlargeuncased\n", 83 | "BertDPRModel\n", 84 | "BertJointIRModel\n", 85 | "DuplexDecoderModel\n", 86 | "neural_text_normalization_t5\n", 87 | "itn_en_t5\n", 88 | "DuplexTaggerModel\n", 89 | "neural_text_normalization_t5\n", 90 | "itn_en_t5\n", 91 | "EntityLinkingModel\n", 92 | "GLUEModel\n", 93 | "IntentSlotClassificationModel\n", 94 | "Joint_Intent_Slot_Assistant\n", 95 | "MTEncDecModel\n", 96 | "nmt_en_de_transformer12x2\n", 97 | "nmt_de_en_transformer12x2\n", 98 | "nmt_en_es_transformer12x2\n", 99 | "nmt_es_en_transformer12x2\n", 100 | "nmt_en_fr_transformer12x2\n", 101 | "nmt_fr_en_transformer12x2\n", 102 | "nmt_en_ru_transformer6x6\n", 103 | "nmt_ru_en_transformer6x6\n", 104 | "nmt_zh_en_transformer6x6\n", 105 | "nmt_en_zh_transformer6x6\n", 106 | "nmt_hi_en_transformer12x2\n", 107 | "nmt_en_hi_transformer12x2\n", 108 | "mnmt_deesfr_en_transformer12x2\n", 109 | "mnmt_deesfr_en_transformer24x6\n", 110 | "mnmt_deesfr_en_transformer6x6\n", 111 | "mnmt_en_deesfr_transformer12x2\n", 112 | "mnmt_en_deesfr_transformer24x6\n", 113 | "mnmt_en_deesfr_transformer6x6\n", 114 | "mnmt_en_deesfr_transformerbase\n", 115 | "nmt_en_de_transformer24x6\n", 116 | "nmt_de_en_transformer24x6\n", 117 | "nmt_en_es_transformer24x6\n", 118 | "nmt_es_en_transformer24x6\n", 119 | "nmt_en_fr_transformer24x6\n", 120 | "nmt_fr_en_transformer24x6\n", 121 | "nmt_en_ru_transformer24x6\n", 122 | "nmt_ru_en_transformer24x6\n", 123 | "nmt_en_zh_transformer24x6\n", 124 | "nmt_zh_en_transformer24x6\n", 125 | "MegatronGPTPromptLearningModel\n", 126 | "MultiLabelIntentSlotClassificationModel\n", 127 | "PunctuationCapitalizationLexicalAudioModel\n", 128 | "PunctuationCapitalizationModel\n", 129 | "punctuation_en_bert\n", 130 | "punctuation_en_distilbert\n", 131 | "QAModel\n", 132 | "qa_squadv1.1_bertbase\n", 133 | "qa_squadv2.0_bertbase\n", 134 | "qa_squadv1_1_bertlarge\n", 135 | "qa_squadv2.0_bertlarge\n", 136 | "qa_squadv1_1_megatron_cased\n", 137 | "qa_squadv2.0_megatron_cased\n", 138 | "qa_squadv1.1_megatron_uncased\n", 139 | "qa_squadv2.0_megatron_uncased\n", 140 | "SpellcheckingAsrCustomizationModel\n", 141 | "Text2SparqlModel\n", 142 | "TextClassificationModel\n", 143 | "ThutmoseTaggerModel\n", 144 | "itn_en_thutmose_bert\n", 145 | "itn_ru_thutmose_bert\n", 146 | "TokenClassificationModel\n", 147 | "ner_en_bert\n", 148 | "TransformerLMModel\n", 149 | "ZeroShotIntentModel\n", 150 | "zeroshotintent_en_bert_base_uncased\n", 151 | "zeroshotintent_en_megatron_uncased\n" 152 | ] 153 | } 154 | ], 155 | "source": [ 156 | "\n", 157 | "all_models = {}\n", 158 | "for mdo_name in nlp_base_models:\n", 159 | " # Get the attribute itself \n", 160 | " mdo = getattr(nemo_nlp.models, mdo_name) \n", 161 | " \n", 162 | " # Check if the attribute is callable and has the 'play' method \n", 163 | " if callable(mdo) and hasattr(mdo, 'list_available_models'): \n", 164 | " play_method = getattr(mdo, 'list_available_models') \n", 165 | " \n", 166 | " # Ensure 'play' is callable before calling it \n", 167 | " if callable(play_method): \n", 168 | " print(mdo_name)\n", 169 | " all_models[mdo_name] = play_method()\n", 170 | "\n", 171 | " if play_method() is not None:\n", 172 | " for one_model in play_method():\n", 173 | " print(one_model.pretrained_model_name)\n", 174 | " # try:\n", 175 | " # model = mdo.from_pretrained(one_model.pretrained_model_name)\n", 176 | " # if not os.path.exists(mdo_name):\n", 177 | " # os.mkdir(mdo_name)\n", 178 | " # model.save_to(mdo_name + '/' + one_model.pretrained_model_name + '.nemo')\n", 179 | " # except Exception as e:\n", 180 | " # print('error:' , e)" 181 | ], 182 | "metadata": { 183 | "collapsed": false, 184 | "ExecuteTime": { 185 | "end_time": "2024-11-01T04:26:59.943212600Z", 186 | "start_time": "2024-11-01T04:26:59.927259500Z" 187 | } 188 | }, 189 | "id": "7f897202876547b7" 190 | }, 191 | { 192 | "cell_type": "code", 193 | "execution_count": 54, 194 | "outputs": [ 195 | { 196 | "name": "stdout", 197 | "output_type": "stream", 198 | "text": [ 199 | "[PretrainedModelInfo(\n", 200 | "\tpretrained_model_name=bertbaseuncased,\n", 201 | "\tdescription=The model was trained EN Wikipedia and BookCorpus on a sequence length of 512.,\n", 202 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/bertbaseuncased/versions/1.0.0rc1/files/bertbaseuncased.nemo\n", 203 | "), PretrainedModelInfo(\n", 204 | "\tpretrained_model_name=bertlargeuncased,\n", 205 | "\tdescription=The model was trained EN Wikipedia and BookCorpus on a sequence length of 512.,\n", 206 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/bertlargeuncased/versions/1.0.0rc1/files/bertlargeuncased.nemo\n", 207 | ")]\n", 208 | "[PretrainedModelInfo(\n", 209 | "\tpretrained_model_name=neural_text_normalization_t5,\n", 210 | "\tdescription=Text Normalization model's decoder model.,\n", 211 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/neural_text_normalization_t5/versions/1.5.0/files/neural_text_normalization_t5_decoder.nemo\n", 212 | "), PretrainedModelInfo(\n", 213 | "\tpretrained_model_name=itn_en_t5,\n", 214 | "\tdescription=English Inverse Text Normalization model's decoder model.,\n", 215 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_en_t5/versions/1.11.0/files/itn_en_t5_decoder.nemo\n", 216 | ")]\n", 217 | "[PretrainedModelInfo(\n", 218 | "\tpretrained_model_name=neural_text_normalization_t5,\n", 219 | "\tdescription=Text Normalization model's tagger model.,\n", 220 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/neural_text_normalization_t5/versions/1.5.0/files/neural_text_normalization_t5_tagger.nemo\n", 221 | "), PretrainedModelInfo(\n", 222 | "\tpretrained_model_name=itn_en_t5,\n", 223 | "\tdescription=English Inverse Text Normalization model's tagger model.,\n", 224 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_en_t5/versions/1.11.0/files/itn_en_t5_tagger.nemo\n", 225 | ")]\n", 226 | "[PretrainedModelInfo(\n", 227 | "\tpretrained_model_name=Joint_Intent_Slot_Assistant,\n", 228 | "\tdescription=This models is trained on this https://github.com/xliuhw/NLU-Evaluation-Data dataset which includes 64 various intents and 55 slots. Final Intent accuracy is about 87%, Slot accuracy is about 89%.,\n", 229 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemonlpmodels/versions/1.0.0a5/files/Joint_Intent_Slot_Assistant.nemo\n", 230 | ")]\n", 231 | "[PretrainedModelInfo(\n", 232 | "\tpretrained_model_name=nmt_en_de_transformer12x2,\n", 233 | "\tdescription=En->De translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_de_transformer12x2,\n", 234 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_de_transformer12x2/versions/1.0.0rc1/files/nmt_en_de_transformer12x2.nemo\n", 235 | "), PretrainedModelInfo(\n", 236 | "\tpretrained_model_name=nmt_de_en_transformer12x2,\n", 237 | "\tdescription=De->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_de_en_transformer12x2,\n", 238 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_de_en_transformer12x2/versions/1.0.0rc1/files/nmt_de_en_transformer12x2.nemo\n", 239 | "), PretrainedModelInfo(\n", 240 | "\tpretrained_model_name=nmt_en_es_transformer12x2,\n", 241 | "\tdescription=En->Es translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_es_transformer12x2,\n", 242 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_es_transformer12x2/versions/1.0.0rc1/files/nmt_en_es_transformer12x2.nemo\n", 243 | "), PretrainedModelInfo(\n", 244 | "\tpretrained_model_name=nmt_es_en_transformer12x2,\n", 245 | "\tdescription=Es->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_es_en_transformer12x2,\n", 246 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_es_en_transformer12x2/versions/1.0.0rc1/files/nmt_es_en_transformer12x2.nemo\n", 247 | "), PretrainedModelInfo(\n", 248 | "\tpretrained_model_name=nmt_en_fr_transformer12x2,\n", 249 | "\tdescription=En->Fr translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_fr_transformer12x2,\n", 250 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_fr_transformer12x2/versions/1.0.0rc1/files/nmt_en_fr_transformer12x2.nemo\n", 251 | "), PretrainedModelInfo(\n", 252 | "\tpretrained_model_name=nmt_fr_en_transformer12x2,\n", 253 | "\tdescription=Fr->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_fr_en_transformer12x2,\n", 254 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_fr_en_transformer12x2/versions/1.0.0rc1/files/nmt_fr_en_transformer12x2.nemo\n", 255 | "), PretrainedModelInfo(\n", 256 | "\tpretrained_model_name=nmt_en_ru_transformer6x6,\n", 257 | "\tdescription=En->Ru translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_ru_transformer6x6,\n", 258 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_ru_transformer6x6/versions/1.0.0rc1/files/nmt_en_ru_transformer6x6.nemo\n", 259 | "), PretrainedModelInfo(\n", 260 | "\tpretrained_model_name=nmt_ru_en_transformer6x6,\n", 261 | "\tdescription=Ru->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_ru_en_transformer6x6,\n", 262 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_ru_en_transformer6x6/versions/1.0.0rc1/files/nmt_ru_en_transformer6x6.nemo\n", 263 | "), PretrainedModelInfo(\n", 264 | "\tpretrained_model_name=nmt_zh_en_transformer6x6,\n", 265 | "\tdescription=Zh->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_zh_en_transformer6x6,\n", 266 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_zh_en_transformer6x6/versions/1.0.0rc1/files/nmt_zh_en_transformer6x6.nemo\n", 267 | "), PretrainedModelInfo(\n", 268 | "\tpretrained_model_name=nmt_en_zh_transformer6x6,\n", 269 | "\tdescription=En->Zh translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_zh_transformer6x6,\n", 270 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_zh_transformer6x6/versions/1.0.0rc1/files/nmt_en_zh_transformer6x6.nemo\n", 271 | "), PretrainedModelInfo(\n", 272 | "\tpretrained_model_name=nmt_hi_en_transformer12x2,\n", 273 | "\tdescription=Hi->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_hi_en_transformer12x2,\n", 274 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_hi_en_transformer12x2/versions/v1.0.0/files/nmt_hi_en_transformer12x2.nemo\n", 275 | "), PretrainedModelInfo(\n", 276 | "\tpretrained_model_name=nmt_en_hi_transformer12x2,\n", 277 | "\tdescription=En->Hi translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_hi_transformer12x2,\n", 278 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_hi_transformer12x2/versions/v1.0.0/files/nmt_en_hi_transformer12x2.nemo\n", 279 | "), PretrainedModelInfo(\n", 280 | "\tpretrained_model_name=mnmt_deesfr_en_transformer12x2,\n", 281 | "\tdescription=De/Es/Fr->En multilingual many-one translation model. The model has 12 encoder and 2 decoder layers with hidden dim 1,024. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_deesfr_en_transformer12x2,\n", 282 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_deesfr_en_transformer12x2/versions/1.2.0/files/mnmt_deesfr_en_transformer12x2.nemo\n", 283 | "), PretrainedModelInfo(\n", 284 | "\tpretrained_model_name=mnmt_deesfr_en_transformer24x6,\n", 285 | "\tdescription=De/Es/Fr->En multilingual many-one translation model. The model has 24 encoder and 6 decoder layers with hidden dim 1,024. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_deesfr_en_transformer24x6,\n", 286 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_deesfr_en_transformer24x6/versions/1.2.0/files/mnmt_deesfr_en_transformer24x6.nemo\n", 287 | "), PretrainedModelInfo(\n", 288 | "\tpretrained_model_name=mnmt_deesfr_en_transformer6x6,\n", 289 | "\tdescription=De/Es/Fr->En multilingual many-one translation model. The model has 6 encoder and 6 decoder layers with hidden dim 1,024. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_deesfr_en_transformer6x6,\n", 290 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_deesfr_en_transformer6x6/versions/1.2.0/files/mnmt_deesfr_en_transformer6x6.nemo\n", 291 | "), PretrainedModelInfo(\n", 292 | "\tpretrained_model_name=mnmt_en_deesfr_transformer12x2,\n", 293 | "\tdescription=En->De/Es/Fr multilingual one-many translation model. The model has 12 encoder and 2 decoder layers with hidden dim 1,024. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_en_deesfr_transformer12x2,\n", 294 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformer12x2/versions/1.2.0/files/mnmt_en_deesfr_transformer12x2.nemo\n", 295 | "), PretrainedModelInfo(\n", 296 | "\tpretrained_model_name=mnmt_en_deesfr_transformer24x6,\n", 297 | "\tdescription=En->De/Es/Fr multilingual one-many translation model. The model has 24 encoder and 6 decoder layers with hidden dim 1,024. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_en_deesfr_transformer24x6,\n", 298 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformer24x6/versions/1.2.0/files/mnmt_en_deesfr_transformer24x6.nemo\n", 299 | "), PretrainedModelInfo(\n", 300 | "\tpretrained_model_name=mnmt_en_deesfr_transformer6x6,\n", 301 | "\tdescription=En->De/Es/Fr multilingual one-many translation model. The model has 6 encoder and 6 decoder layers with hidden dim 1,024. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_en_deesfr_transformer6x6,\n", 302 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformer6x6/versions/1.2.0/files/mnmt_en_deesfr_transformer6x6.nemo\n", 303 | "), PretrainedModelInfo(\n", 304 | "\tpretrained_model_name=mnmt_en_deesfr_transformerbase,\n", 305 | "\tdescription=En->De/Es/Fr multilingual one-many translation model. The model has 6 encoder and 6 decoder layers with hidden dim 512. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:mnmt_en_deesfr_transformerbase,\n", 306 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/mnmt_en_deesfr_transformerbase/versions/1.2.0/files/mnmt_en_deesfr_transformerbase.nemo\n", 307 | "), PretrainedModelInfo(\n", 308 | "\tpretrained_model_name=nmt_en_de_transformer24x6,\n", 309 | "\tdescription=En->De translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_de_transformer24x6,\n", 310 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_de_transformer24x6/versions/1.5/files/en_de_24x6.nemo\n", 311 | "), PretrainedModelInfo(\n", 312 | "\tpretrained_model_name=nmt_de_en_transformer24x6,\n", 313 | "\tdescription=De->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_de_en_transformer24x6,\n", 314 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_de_en_transformer24x6/versions/1.5/files/de_en_24x6.nemo\n", 315 | "), PretrainedModelInfo(\n", 316 | "\tpretrained_model_name=nmt_en_es_transformer24x6,\n", 317 | "\tdescription=En->Es translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_es_transformer24x6,\n", 318 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_es_transformer24x6/versions/1.5/files/en_es_24x6.nemo\n", 319 | "), PretrainedModelInfo(\n", 320 | "\tpretrained_model_name=nmt_es_en_transformer24x6,\n", 321 | "\tdescription=Es->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_es_en_transformer24x6,\n", 322 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_es_en_transformer24x6/versions/1.5/files/es_en_24x6.nemo\n", 323 | "), PretrainedModelInfo(\n", 324 | "\tpretrained_model_name=nmt_en_fr_transformer24x6,\n", 325 | "\tdescription=En->Fr translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_fr_transformer24x6,\n", 326 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_fr_transformer24x6/versions/1.5/files/en_fr_24x6.nemo\n", 327 | "), PretrainedModelInfo(\n", 328 | "\tpretrained_model_name=nmt_fr_en_transformer24x6,\n", 329 | "\tdescription=Fr->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_fr_en_transformer24x6,\n", 330 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_fr_en_transformer24x6/versions/1.5/files/fr_en_24x6.nemo\n", 331 | "), PretrainedModelInfo(\n", 332 | "\tpretrained_model_name=nmt_en_ru_transformer24x6,\n", 333 | "\tdescription=En->Ru translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_ru_transformer24x6,\n", 334 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_ru_transformer24x6/versions/1.5/files/en_ru_24x6.nemo\n", 335 | "), PretrainedModelInfo(\n", 336 | "\tpretrained_model_name=nmt_ru_en_transformer24x6,\n", 337 | "\tdescription=Ru->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_ru_en_transformer24x6,\n", 338 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_ru_en_transformer24x6/versions/1.5/files/ru_en_24x6.nemo\n", 339 | "), PretrainedModelInfo(\n", 340 | "\tpretrained_model_name=nmt_en_zh_transformer24x6,\n", 341 | "\tdescription=En->Zh translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_en_zh_transformer24x6,\n", 342 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_en_zh_transformer24x6/versions/1.5/files/en_zh_24x6.nemo\n", 343 | "), PretrainedModelInfo(\n", 344 | "\tpretrained_model_name=nmt_zh_en_transformer24x6,\n", 345 | "\tdescription=Zh->En translation model. See details here: https://ngc.nvidia.com/catalog/models/nvidia:nemo:nmt_zh_en_transformer24x6,\n", 346 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/nmt_zh_en_transformer24x6/versions/1.5/files/zh_en_24x6.nemo\n", 347 | ")]\n", 348 | "[]\n", 349 | "[]\n", 350 | "[PretrainedModelInfo(\n", 351 | "\tpretrained_model_name=punctuation_en_bert,\n", 352 | "\tdescription=The model was trained with NeMo BERT base uncased checkpoint on a subset of data from the following sources: Tatoeba sentences, books from Project Gutenberg, Fisher transcripts.,\n", 353 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/punctuation_en_bert/versions/1.0.0rc1/files/punctuation_en_bert.nemo\n", 354 | "), PretrainedModelInfo(\n", 355 | "\tpretrained_model_name=punctuation_en_distilbert,\n", 356 | "\tdescription=The model was trained with DistilBERT base uncased checkpoint from HuggingFace on a subset of data from the following sources: Tatoeba sentences, books from Project Gutenberg, Fisher transcripts.,\n", 357 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/punctuation_en_distilbert/versions/1.0.0rc1/files/punctuation_en_distilbert.nemo\n", 358 | ")]\n", 359 | "[PretrainedModelInfo(\n", 360 | "\tpretrained_model_name=qa_squadv1.1_bertbase,\n", 361 | "\tdescription=Question answering model finetuned from NeMo BERT Base Uncased on SQuAD v1.1 dataset which obtains an exact match (EM) score of 82.78% and an F1 score of 89.97%.,\n", 362 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_bertbase/versions/1.0.0rc1/files/qa_squadv1.1_bertbase.nemo\n", 363 | "), PretrainedModelInfo(\n", 364 | "\tpretrained_model_name=qa_squadv2.0_bertbase,\n", 365 | "\tdescription=Question answering model finetuned from NeMo BERT Base Uncased on SQuAD v2.0 dataset which obtains an exact match (EM) score of 75.04% and an F1 score of 78.08%.,\n", 366 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_bertbase/versions/1.0.0rc1/files/qa_squadv2.0_bertbase.nemo\n", 367 | "), PretrainedModelInfo(\n", 368 | "\tpretrained_model_name=qa_squadv1_1_bertlarge,\n", 369 | "\tdescription=Question answering model finetuned from NeMo BERT Large Uncased on SQuAD v1.1 dataset which obtains an exact match (EM) score of 85.44% and an F1 score of 92.06%.,\n", 370 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_bertlarge/versions/1.0.0rc1/files/qa_squadv1.1_bertlarge.nemo\n", 371 | "), PretrainedModelInfo(\n", 372 | "\tpretrained_model_name=qa_squadv2.0_bertlarge,\n", 373 | "\tdescription=Question answering model finetuned from NeMo BERT Large Uncased on SQuAD v2.0 dataset which obtains an exact match (EM) score of 80.22% and an F1 score of 83.05%.,\n", 374 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_bertlarge/versions/1.0.0rc1/files/qa_squadv2.0_bertlarge.nemo\n", 375 | "), PretrainedModelInfo(\n", 376 | "\tpretrained_model_name=qa_squadv1_1_megatron_cased,\n", 377 | "\tdescription=Question answering model finetuned from Megatron Cased on SQuAD v1.1 dataset which obtains an exact match (EM) score of 88.18% and an F1 score of 94.07%.,\n", 378 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_megatron_cased/versions/1.0.0rc1/files/qa_squadv1.1_megatron_cased.nemo\n", 379 | "), PretrainedModelInfo(\n", 380 | "\tpretrained_model_name=qa_squadv2.0_megatron_cased,\n", 381 | "\tdescription=Question answering model finetuned from Megatron Cased on SQuAD v2.0 dataset which obtains an exact match (EM) score of 84.73% and an F1 score of 87.89%.,\n", 382 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_megatron_cased/versions/1.0.0rc1/files/qa_squadv2.0_megatron_cased.nemo\n", 383 | "), PretrainedModelInfo(\n", 384 | "\tpretrained_model_name=qa_squadv1.1_megatron_uncased,\n", 385 | "\tdescription=Question answering model finetuned from Megatron Unased on SQuAD v1.1 dataset which obtains an exact match (EM) score of 87.61% and an F1 score of 94.00%.,\n", 386 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv1_1_megatron_uncased/versions/1.0.0rc1/files/qa_squadv1.1_megatron_uncased.nemo\n", 387 | "), PretrainedModelInfo(\n", 388 | "\tpretrained_model_name=qa_squadv2.0_megatron_uncased,\n", 389 | "\tdescription=Question answering model finetuned from Megatron Uncased on SQuAD v2.0 dataset which obtains an exact match (EM) score of 84.48% and an F1 score of 87.65%.,\n", 390 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/qa_squadv2_0_megatron_uncased/versions/1.0.0rc1/files/qa_squadv2.0_megatron_uncased.nemo\n", 391 | ")]\n", 392 | "[PretrainedModelInfo(\n", 393 | "\tpretrained_model_name=itn_en_thutmose_bert,\n", 394 | "\tdescription=A single-pass tagger-based English model for inverse text normalization basedon BERT, trained on 2 mln sentences from Google Text Normalization Dataset,\n", 395 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_en_thutmose_bert/versions/1.9.0/files/itn_en_thutmose_bert.nemo\n", 396 | "), PretrainedModelInfo(\n", 397 | "\tpretrained_model_name=itn_ru_thutmose_bert,\n", 398 | "\tdescription=A single-pass tagger-based Russian model for inverse text normalization basedon BERT, trained on 2 mln sentences from Google Text Normalization Dataset,\n", 399 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/itn_ru_thutmose_bert/versions/1.11.0/files/itn_ru_thutmose_bert.nemo\n", 400 | ")]\n", 401 | "[PretrainedModelInfo(\n", 402 | "\tpretrained_model_name=ner_en_bert,\n", 403 | "\tdescription=The model was trained on GMB (Groningen Meaning Bank) corpus for entity recognition and achieves 74.61 F1 Macro score.,\n", 404 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/ner_en_bert/versions/1.10/files/ner_en_bert.nemo\n", 405 | ")]\n", 406 | "[PretrainedModelInfo(\n", 407 | "\tpretrained_model_name=zeroshotintent_en_bert_base_uncased,\n", 408 | "\tdescription=ZeroShotIntentModel trained by fine tuning BERT-base-uncased on the MNLI (Multi-Genre Natural Language Inference) dataset, which achieves an accuracy of 84.9% and 84.8% on the matched and mismatched dev sets, respectively.,\n", 409 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/zeroshotintent_en_bert_base_uncased/versions/1.4.1/files/zeroshotintent_en_bert_base_uncased.nemo\n", 410 | "), PretrainedModelInfo(\n", 411 | "\tpretrained_model_name=zeroshotintent_en_megatron_uncased,\n", 412 | "\tdescription=ZeroShotIntentModel trained by fine tuning Megatron-BERT-345m=M-uncased on the MNLI (Multi-Genre Natural Language Inference) dataset, which achieves an accuracy of 90.0% and 89.9% on the matched and mismatched dev sets, respectively,\n", 413 | "\tlocation=https://api.ngc.nvidia.com/v2/models/nvidia/nemo/zeroshotintent_en_megatron_uncased/versions/1.4.1/files/zeroshotintent_en_megatron_uncased.nemo\n", 414 | ")]\n" 415 | ] 416 | } 417 | ], 418 | "source": [ 419 | "\n", 420 | "for model in all_models.keys():\n", 421 | " if all_models[model] is not None:\n", 422 | " print(all_models[model])\n", 423 | " for each_model in all_models[model]:\n", 424 | " with open('download_wget_nlp.txt', 'a') as file:\n", 425 | " if each_model.location is not None:\n", 426 | " file.write(each_model.location + '\\n')" 427 | ], 428 | "metadata": { 429 | "collapsed": false, 430 | "ExecuteTime": { 431 | "end_time": "2024-11-01T04:27:41.309497400Z", 432 | "start_time": "2024-11-01T04:27:41.278194400Z" 433 | } 434 | }, 435 | "id": "2c865927ceaaa279" 436 | }, 437 | { 438 | "cell_type": "code", 439 | "execution_count": 30, 440 | "outputs": [ 441 | { 442 | "name": "stdout", 443 | "output_type": "stream", 444 | "text": [ 445 | "[NeMo I 2024-11-01 04:00:27 cloud:68] Downloading from: https://api.ngc.nvidia.com/v2/models/nvidia/nemospeechmodels/versions/1.0.0a5/files/QuartzNet15x5Base-En.nemo to C:\\Users\\Administrator\\.cache\\torch\\NeMo\\NeMo_1.21.0\\QuartzNet15x5Base-En\\2b066be39e9294d7100fb176ec817722\\QuartzNet15x5Base-En.nemo\n", 446 | "[NeMo I 2024-11-01 04:00:28 common:913] Instantiating model from pre-trained checkpoint\n", 447 | "[NeMo I 2024-11-01 04:00:29 features:289] PADDING: 16\n", 448 | "[NeMo I 2024-11-01 04:00:30 save_restore_connector:249] Model EncDecCTCModel was successfully restored from C:\\Users\\Administrator\\.cache\\torch\\NeMo\\NeMo_1.21.0\\QuartzNet15x5Base-En\\2b066be39e9294d7100fb176ec817722\\QuartzNet15x5Base-En.nemo.\n" 449 | ] 450 | }, 451 | { 452 | "data": { 453 | "text/plain": "EncDecCTCModel(\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (encoder): ConvASREncoder(\n (encoder): Sequential(\n (0): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(33,), stride=(2,), padding=(16,), groups=64, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(64, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (1): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (2): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (3): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(33,), stride=(1,), padding=(16,), groups=256, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (4): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (5): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (6): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(39,), stride=(1,), padding=(19,), groups=256, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(256, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (7): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 256, kernel_size=(51,), stride=(1,), padding=(25,), groups=256, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(256, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(256, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (8): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (9): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(51,), stride=(1,), padding=(25,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (10): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (11): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (12): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(63,), stride=(1,), padding=(31,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (13): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (14): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (15): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.0, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): ReLU(inplace=True)\n (14): Dropout(p=0.0, inplace=False)\n (15): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (16): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (17): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (18): ReLU(inplace=True)\n (19): Dropout(p=0.0, inplace=False)\n (20): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(75,), stride=(1,), padding=(37,), groups=512, bias=False)\n )\n (21): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (22): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (16): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(87,), stride=(1,), padding=(86,), dilation=(2,), groups=512, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(512, 512, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(512, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (17): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(512, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n )\n )\n (decoder): ConvASRDecoder(\n (decoder_layers): Sequential(\n (0): Conv1d(1024, 29, kernel_size=(1,), stride=(1,))\n )\n )\n (loss): CTCLoss()\n (spec_augmentation): SpectrogramAugmentation(\n (spec_cutout): SpecCutout()\n )\n (_wer): WER()\n)" 454 | }, 455 | "execution_count": 30, 456 | "metadata": {}, 457 | "output_type": "execute_result" 458 | } 459 | ], 460 | "source": [ 461 | "nemo_asr.models.ASRModel.from_pretrained('QuartzNet15x5Base-En')\n" 462 | ], 463 | "metadata": { 464 | "collapsed": false, 465 | "ExecuteTime": { 466 | "end_time": "2024-11-01T04:00:30.544724Z", 467 | "start_time": "2024-11-01T04:00:27.480559Z" 468 | } 469 | }, 470 | "id": "1018d9c1a387f00b" 471 | }, 472 | { 473 | "cell_type": "code", 474 | "execution_count": 55, 475 | "outputs": [], 476 | "source": [ 477 | "import wget" 478 | ], 479 | "metadata": { 480 | "collapsed": false, 481 | "ExecuteTime": { 482 | "end_time": "2024-11-01T04:34:23.151042200Z", 483 | "start_time": "2024-11-01T04:34:23.119165500Z" 484 | } 485 | }, 486 | "id": "960d51ff6bb6b682" 487 | }, 488 | { 489 | "cell_type": "code", 490 | "execution_count": 57, 491 | "outputs": [ 492 | { 493 | "data": { 494 | "text/plain": "'ssl_en (1).nemo'" 495 | }, 496 | "execution_count": 57, 497 | "metadata": {}, 498 | "output_type": "execute_result" 499 | } 500 | ], 501 | "source": [ 502 | "wget.download('https://api.ngc.nvidia.com/v2/models/nvidia/nemospeechmodels/versions/1.0.0a5/files/QuartzNet15x5Base-En.nemo', 'ssl_en.nemo')\n" 503 | ], 504 | "metadata": { 505 | "collapsed": false, 506 | "ExecuteTime": { 507 | "end_time": "2024-11-01T04:37:28.879382Z", 508 | "start_time": "2024-11-01T04:37:27.492911200Z" 509 | } 510 | }, 511 | "id": "1a0ce11417a4c2a0" 512 | }, 513 | { 514 | "cell_type": "code", 515 | "execution_count": 7, 516 | "outputs": [ 517 | { 518 | "name": "stdout", 519 | "output_type": "stream", 520 | "text": [ 521 | "QuartzNet15x5Base-En\n", 522 | "asr_talknet_aligner\n", 523 | "commandrecognition_en_matchboxnet3x1x64_v1\n", 524 | "commandrecognition_en_matchboxnet3x1x64_v2\n", 525 | "commandrecognition_en_matchboxnet3x1x64_v2_subset_task\n", 526 | "commandrecognition_en_matchboxnet3x2x64_v1\n", 527 | "commandrecognition_en_matchboxnet3x2x64_v2\n", 528 | "commandrecognition_en_matchboxnet3x2x64_v2_subset_task\n", 529 | "stt_be_conformer_ctc_large\n", 530 | "stt_be_conformer_transducer_large\n", 531 | "stt_by_fastconformer_hybrid_large_pc\n", 532 | "stt_ca_conformer_ctc_large\n", 533 | "stt_ca_conformer_transducer_large\n", 534 | "stt_ca_quartznet15x5\n", 535 | "stt_de_citrinet_1024\n", 536 | "stt_de_conformer_ctc_large\n", 537 | "stt_de_conformer_transducer_large\n", 538 | "stt_de_contextnet_1024\n", 539 | "stt_de_fastconformer_hybrid_large_pc\n", 540 | "stt_de_quartznet15x5\n", 541 | "stt_en_citrinet_1024\n", 542 | "stt_en_citrinet_1024_gamma_0_25\n", 543 | "stt_en_citrinet_256\n", 544 | "stt_en_citrinet_256_gamma_0_25\n", 545 | "stt_en_citrinet_512\n", 546 | "stt_en_citrinet_512_gamma_0_25\n", 547 | "stt_en_conformer_ctc_large\n", 548 | "stt_en_conformer_ctc_large_ls\n", 549 | "stt_en_conformer_ctc_medium\n", 550 | "stt_en_conformer_ctc_medium_ls\n", 551 | "stt_en_conformer_ctc_small\n", 552 | "stt_en_conformer_ctc_small_ls\n", 553 | "stt_en_conformer_ctc_xlarge\n", 554 | "stt_en_conformer_transducer_large\n", 555 | "stt_en_conformer_transducer_large_ls\n", 556 | "stt_en_conformer_transducer_medium\n", 557 | "stt_en_conformer_transducer_small\n", 558 | "stt_en_conformer_transducer_xlarge\n", 559 | "stt_en_conformer_transducer_xxlarge\n", 560 | "stt_en_contextnet_1024\n", 561 | "stt_en_contextnet_1024_mls\n", 562 | "stt_en_contextnet_256\n", 563 | "stt_en_contextnet_256_mls\n", 564 | "stt_en_contextnet_512\n", 565 | "stt_en_contextnet_512_mls\n", 566 | "stt_en_fastconformer_ctc_large\n", 567 | "stt_en_fastconformer_ctc_xlarge\n", 568 | "stt_en_fastconformer_ctc_xxlarge\n", 569 | "stt_en_fastconformer_hybrid_large_pc\n", 570 | "stt_en_fastconformer_hybrid_large_streaming_1040ms\n", 571 | "stt_en_fastconformer_hybrid_large_streaming_480ms\n", 572 | "stt_en_fastconformer_hybrid_large_streaming_80ms\n", 573 | "stt_en_fastconformer_hybrid_large_streaming_multi\n", 574 | "stt_en_fastconformer_transducer_large\n", 575 | "stt_en_fastconformer_transducer_xlarge\n", 576 | "stt_en_fastconformer_transducer_xxlarge\n", 577 | "stt_en_jasper10x5dr\n", 578 | "stt_en_quartznet15x5\n", 579 | "stt_en_squeezeformer_ctc_large_ls\n", 580 | "stt_en_squeezeformer_ctc_medium_large_ls\n", 581 | "stt_en_squeezeformer_ctc_medium_ls\n", 582 | "stt_en_squeezeformer_ctc_small_ls\n", 583 | "stt_en_squeezeformer_ctc_small_medium_ls\n", 584 | "stt_en_squeezeformer_ctc_xsmall_ls\n", 585 | "stt_enes_conformer_ctc_large\n", 586 | "stt_enes_conformer_ctc_large_codesw\n", 587 | "stt_enes_conformer_transducer_large\n", 588 | "stt_enes_conformer_transducer_large_codesw\n", 589 | "stt_enes_contextnet_large\n", 590 | "stt_eo_conformer_ctc_large\n", 591 | "stt_eo_conformer_transducer_large\n", 592 | "stt_es_citrinet_1024_gamma_0_25\n", 593 | "stt_es_citrinet_512\n", 594 | "stt_es_conformer_ctc_large\n", 595 | "stt_es_conformer_transducer_large\n", 596 | "stt_es_contextnet_1024\n", 597 | "stt_es_fastconformer_hybrid_large_pc\n", 598 | "stt_es_quartznet15x5\n", 599 | "stt_fr_citrinet_1024_gamma_0_25\n", 600 | "stt_fr_conformer_ctc_large\n", 601 | "stt_fr_conformer_transducer_large\n", 602 | "stt_fr_contextnet_1024\n", 603 | "stt_fr_fastconformer_hybrid_large_pc\n", 604 | "stt_fr_no_hyphen_citrinet_1024_gamma_0_25\n", 605 | "stt_fr_no_hyphen_conformer_ctc_large\n", 606 | "stt_fr_quartznet15x5\n", 607 | "stt_hi_conformer_ctc_medium\n", 608 | "stt_hr_conformer_ctc_large\n", 609 | "stt_hr_conformer_transducer_large\n", 610 | "stt_hr_fastconformer_hybrid_large_pc\n", 611 | "stt_it_conformer_ctc_large\n", 612 | "stt_it_conformer_transducer_large\n", 613 | "stt_it_fastconformer_hybrid_large_pc\n", 614 | "stt_it_quartznet15x5\n", 615 | "stt_kab_conformer_transducer_large\n", 616 | "stt_mr_conformer_ctc_medium\n", 617 | "stt_multilingual_fastconformer_hybrid_large_pc\n", 618 | "stt_multilingual_fastconformer_hybrid_large_pc_blend_eu\n", 619 | "stt_pl_fastconformer_hybrid_large_pc\n", 620 | "stt_pl_quartznet15x5\n", 621 | "stt_ru_conformer_ctc_large\n", 622 | "stt_ru_conformer_transducer_large\n", 623 | "stt_ru_fastconformer_hybrid_large_pc\n", 624 | "stt_ru_quartznet15x5\n", 625 | "stt_rw_conformer_ctc_large\n", 626 | "stt_rw_conformer_transducer_large\n", 627 | "stt_ua_fastconformer_hybrid_large_pc\n", 628 | "stt_zh_citrinet_1024_gamma_0_25\n", 629 | "stt_zh_citrinet_512\n", 630 | "stt_zh_conformer_transducer_large\n", 631 | "vad_marblenet\n", 632 | "vad_multilingual_frame_marblenet\n", 633 | "vad_multilingual_marblenet\n", 634 | "vad_telephony_marblenet\n" 635 | ] 636 | } 637 | ], 638 | "source": [ 639 | "for pretrained_model in all_models['ASRModel']:\n", 640 | " print(pretrained_model.pretrained_model_name)" 641 | ], 642 | "metadata": { 643 | "collapsed": false, 644 | "ExecuteTime": { 645 | "end_time": "2024-10-31T22:55:12.161291100Z", 646 | "start_time": "2024-10-31T22:55:12.144418900Z" 647 | } 648 | }, 649 | "id": "de70c1cb94200670" 650 | }, 651 | { 652 | "cell_type": "code", 653 | "execution_count": 22, 654 | "outputs": [ 655 | { 656 | "name": "stderr", 657 | "output_type": "stream", 658 | "text": [ 659 | "[NeMo W 2024-10-31 19:45:57 modelPT:161] If you intend to do training or fine-tuning, please call the ModelPT.setup_training_data() method and provide a valid configuration file to setup the train data loader.\n", 660 | " Train config : \n", 661 | " manifest_filepath: null\n", 662 | " emb_dir: null\n", 663 | " sample_rate: 16000\n", 664 | " num_spks: 2\n", 665 | " soft_label_thres: 0.5\n", 666 | " labels: null\n", 667 | " batch_size: 15\n", 668 | " emb_batch_size: 0\n", 669 | " shuffle: true\n", 670 | " \n", 671 | "[NeMo W 2024-10-31 19:45:57 modelPT:168] If you intend to do validation, please call the ModelPT.setup_validation_data() or ModelPT.setup_multiple_validation_data() method and provide a valid configuration file to setup the validation data loader(s). \n", 672 | " Validation config : \n", 673 | " manifest_filepath: null\n", 674 | " emb_dir: null\n", 675 | " sample_rate: 16000\n", 676 | " num_spks: 2\n", 677 | " soft_label_thres: 0.5\n", 678 | " labels: null\n", 679 | " batch_size: 15\n", 680 | " emb_batch_size: 0\n", 681 | " shuffle: false\n", 682 | " \n", 683 | "[NeMo W 2024-10-31 19:45:57 modelPT:174] Please call the ModelPT.setup_test_data() or ModelPT.setup_multiple_test_data() method and provide a valid configuration file to setup the test data loader(s).\n", 684 | " Test config : \n", 685 | " manifest_filepath: null\n", 686 | " emb_dir: null\n", 687 | " sample_rate: 16000\n", 688 | " num_spks: 2\n", 689 | " soft_label_thres: 0.5\n", 690 | " labels: null\n", 691 | " batch_size: 15\n", 692 | " emb_batch_size: 0\n", 693 | " shuffle: false\n", 694 | " seq_eval_mode: false\n", 695 | " \n", 696 | "[NeMo W 2024-10-31 19:45:58 nemo_logging:349] C:\\Users\\Administrator\\anaconda3\\envs\\nemo\\lib\\site-packages\\nemo\\core\\connectors\\save_restore_connector.py:568: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n", 697 | " return torch.load(model_weights, map_location='cpu')\n", 698 | " \n", 699 | "[NeMo W 2024-10-31 19:45:59 modelPT:161] If you intend to do training or fine-tuning, please call the ModelPT.setup_training_data() method and provide a valid configuration file to setup the train data loader.\n", 700 | " Train config : \n", 701 | " manifest_filepath: /manifests/ami_train_0.63.json,/manifests/freesound_background_train.json,/manifests/freesound_laughter_train.json,/manifests/fisher_2004_background.json,/manifests/fisher_2004_speech_sampled.json,/manifests/google_train_manifest.json,/manifests/icsi_all_0.63.json,/manifests/musan_freesound_train.json,/manifests/musan_music_train.json,/manifests/musan_soundbible_train.json,/manifests/mandarin_train_sample.json,/manifests/german_train_sample.json,/manifests/spanish_train_sample.json,/manifests/french_train_sample.json,/manifests/russian_train_sample.json\n", 702 | " sample_rate: 16000\n", 703 | " labels:\n", 704 | " - background\n", 705 | " - speech\n", 706 | " batch_size: 256\n", 707 | " shuffle: true\n", 708 | " is_tarred: false\n", 709 | " tarred_audio_filepaths: null\n", 710 | " tarred_shard_strategy: scatter\n", 711 | " augmentor:\n", 712 | " shift:\n", 713 | " prob: 0.5\n", 714 | " min_shift_ms: -10.0\n", 715 | " max_shift_ms: 10.0\n", 716 | " white_noise:\n", 717 | " prob: 0.5\n", 718 | " min_level: -90\n", 719 | " max_level: -46\n", 720 | " norm: true\n", 721 | " noise:\n", 722 | " prob: 0.5\n", 723 | " manifest_path: /manifests/noise_0_1_musan_fs.json\n", 724 | " min_snr_db: 0\n", 725 | " max_snr_db: 30\n", 726 | " max_gain_db: 300.0\n", 727 | " norm: true\n", 728 | " gain:\n", 729 | " prob: 0.5\n", 730 | " min_gain_dbfs: -10.0\n", 731 | " max_gain_dbfs: 10.0\n", 732 | " norm: true\n", 733 | " num_workers: 16\n", 734 | " pin_memory: true\n", 735 | " \n", 736 | "[NeMo W 2024-10-31 19:45:59 modelPT:168] If you intend to do validation, please call the ModelPT.setup_validation_data() or ModelPT.setup_multiple_validation_data() method and provide a valid configuration file to setup the validation data loader(s). \n", 737 | " Validation config : \n", 738 | " manifest_filepath: /manifests/ami_dev_0.63.json,/manifests/freesound_background_dev.json,/manifests/freesound_laughter_dev.json,/manifests/ch120_moved_0.63.json,/manifests/fisher_2005_500_speech_sampled.json,/manifests/google_dev_manifest.json,/manifests/musan_music_dev.json,/manifests/mandarin_dev.json,/manifests/german_dev.json,/manifests/spanish_dev.json,/manifests/french_dev.json,/manifests/russian_dev.json\n", 739 | " sample_rate: 16000\n", 740 | " labels:\n", 741 | " - background\n", 742 | " - speech\n", 743 | " batch_size: 256\n", 744 | " shuffle: false\n", 745 | " val_loss_idx: 0\n", 746 | " num_workers: 16\n", 747 | " pin_memory: true\n", 748 | " \n", 749 | "[NeMo W 2024-10-31 19:45:59 modelPT:174] Please call the ModelPT.setup_test_data() or ModelPT.setup_multiple_test_data() method and provide a valid configuration file to setup the test data loader(s).\n", 750 | " Test config : \n", 751 | " manifest_filepath: null\n", 752 | " sample_rate: 16000\n", 753 | " labels:\n", 754 | " - background\n", 755 | " - speech\n", 756 | " batch_size: 128\n", 757 | " shuffle: false\n", 758 | " test_loss_idx: 0\n", 759 | " \n" 760 | ] 761 | }, 762 | { 763 | "data": { 764 | "text/plain": "NeuralDiarizer(\n (msdd_model): EncDecDiarLabelModel(\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (msdd): MSDD_module(\n (softmax): Softmax(dim=2)\n (cos_dist): CosineSimilarity()\n (lstm): LSTM(256, 256, num_layers=2, batch_first=True, dropout=0.5, bidirectional=True)\n (conv): ModuleList(\n (0): ConvLayer(\n (cnn): Sequential(\n (0): Conv2d(1, 16, kernel_size=(15, 1), stride=(1, 1))\n (1): ReLU()\n (2): BatchNorm2d(16, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n )\n )\n (1): ConvLayer(\n (cnn): Sequential(\n (0): Conv2d(1, 16, kernel_size=(16, 1), stride=(1, 1))\n (1): ReLU()\n (2): BatchNorm2d(16, eps=0.001, momentum=0.99, affine=True, track_running_stats=True)\n )\n )\n )\n (conv_bn): ModuleList(\n (0-1): 2 x BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=False, track_running_stats=True)\n )\n (conv_to_linear): Linear(in_features=3072, out_features=256, bias=True)\n (linear_to_weights): Linear(in_features=256, out_features=5, bias=True)\n (hidden_to_spks): Linear(in_features=512, out_features=2, bias=True)\n (dist_to_emb): Linear(in_features=10, out_features=256, bias=True)\n (dropout): Dropout(p=0.5, inplace=False)\n (_speaker_model): EncDecSpeakerLabelModel(\n (loss): AngularSoftmaxLoss()\n (eval_loss): AngularSoftmaxLoss()\n (_accuracy): TopKClassificationAccuracy()\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (encoder): ConvASREncoder(\n (encoder): Sequential(\n (0): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(80, 80, kernel_size=(3,), stride=(1,), padding=(1,), groups=80, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(80, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (1): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (2): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (3): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (4): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 3072, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(3072, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=3072, out_features=384, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=384, out_features=3072, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n )\n )\n (decoder): SpeakerDecoder(\n (_pooling): AttentivePoolLayer(\n (attention_layer): Sequential(\n (0): TDNNModule(\n (conv_layer): Conv1d(9216, 128, kernel_size=(1,), stride=(1,))\n (activation): ReLU()\n (bn): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n )\n (1): Tanh()\n (2): Conv1d(128, 3072, kernel_size=(1,), stride=(1,))\n )\n )\n (emb_layers): ModuleList(\n (0): Sequential(\n (0): BatchNorm1d(6144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n (1): Conv1d(6144, 192, kernel_size=(1,), stride=(1,))\n )\n )\n (final): Linear(in_features=192, out_features=16681, bias=False)\n )\n (_macro_accuracy): Accuracy()\n (spec_augmentation): SpectrogramAugmentation(\n (spec_augment): SpecAugment()\n )\n )\n )\n (_accuracy_test): MultiBinaryAccuracy()\n (_accuracy_train): MultiBinaryAccuracy()\n (_accuracy_valid): MultiBinaryAccuracy()\n )\n (_speaker_model): EncDecSpeakerLabelModel(\n (loss): AngularSoftmaxLoss()\n (eval_loss): AngularSoftmaxLoss()\n (_accuracy): TopKClassificationAccuracy()\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (encoder): ConvASREncoder(\n (encoder): Sequential(\n (0): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(80, 80, kernel_size=(3,), stride=(1,), padding=(1,), groups=80, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(80, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (1): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (2): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (3): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (4): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 3072, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(3072, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=3072, out_features=384, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=384, out_features=3072, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n )\n )\n (decoder): SpeakerDecoder(\n (_pooling): AttentivePoolLayer(\n (attention_layer): Sequential(\n (0): TDNNModule(\n (conv_layer): Conv1d(9216, 128, kernel_size=(1,), stride=(1,))\n (activation): ReLU()\n (bn): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n )\n (1): Tanh()\n (2): Conv1d(128, 3072, kernel_size=(1,), stride=(1,))\n )\n )\n (emb_layers): ModuleList(\n (0): Sequential(\n (0): BatchNorm1d(6144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n (1): Conv1d(6144, 192, kernel_size=(1,), stride=(1,))\n )\n )\n (final): Linear(in_features=192, out_features=16681, bias=False)\n )\n (_macro_accuracy): Accuracy()\n (spec_augmentation): SpectrogramAugmentation(\n (spec_augment): SpecAugment()\n )\n )\n (clustering_embedding): ClusterEmbedding(\n (_speaker_model): EncDecSpeakerLabelModel(\n (loss): AngularSoftmaxLoss()\n (eval_loss): AngularSoftmaxLoss()\n (_accuracy): TopKClassificationAccuracy()\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (encoder): ConvASREncoder(\n (encoder): Sequential(\n (0): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(80, 80, kernel_size=(3,), stride=(1,), padding=(1,), groups=80, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(80, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (1): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (2): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (3): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (4): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 3072, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(3072, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=3072, out_features=384, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=384, out_features=3072, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n )\n )\n (decoder): SpeakerDecoder(\n (_pooling): AttentivePoolLayer(\n (attention_layer): Sequential(\n (0): TDNNModule(\n (conv_layer): Conv1d(9216, 128, kernel_size=(1,), stride=(1,))\n (activation): ReLU()\n (bn): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n )\n (1): Tanh()\n (2): Conv1d(128, 3072, kernel_size=(1,), stride=(1,))\n )\n )\n (emb_layers): ModuleList(\n (0): Sequential(\n (0): BatchNorm1d(6144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n (1): Conv1d(6144, 192, kernel_size=(1,), stride=(1,))\n )\n )\n (final): Linear(in_features=192, out_features=16681, bias=False)\n )\n (_macro_accuracy): Accuracy()\n (spec_augmentation): SpectrogramAugmentation(\n (spec_augment): SpecAugment()\n )\n )\n (clus_diar_model): ClusteringDiarizer(\n (_vad_model): EncDecClassificationModel(\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (encoder): ConvASREncoder(\n (encoder): Sequential(\n (0): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(80, 80, kernel_size=(11,), stride=(1,), padding=(5,), groups=80, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(80, 128, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (1): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(128, 128, kernel_size=(13,), stride=(1,), padding=(6,), groups=128, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(128, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(13,), stride=(1,), padding=(6,), groups=64, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(128, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (2): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(15,), stride=(1,), padding=(7,), groups=64, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(15,), stride=(1,), padding=(7,), groups=64, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (3): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(17,), stride=(1,), padding=(8,), groups=64, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.0, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(17,), stride=(1,), padding=(8,), groups=64, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(64, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (4): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(64, 64, kernel_size=(29,), stride=(1,), padding=(28,), dilation=(2,), groups=64, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(64, 128, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (5): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(128, 128, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(128, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n )\n )\n (decoder): ConvASRDecoderClassification(\n (pooling): AdaptiveAvgPool1d(output_size=1)\n (decoder_layers): Sequential(\n (0): Linear(in_features=128, out_features=2, bias=True)\n )\n )\n (loss): CrossEntropyLoss()\n (_accuracy): TopKClassificationAccuracy()\n )\n (_speaker_model): EncDecSpeakerLabelModel(\n (loss): AngularSoftmaxLoss()\n (eval_loss): AngularSoftmaxLoss()\n (_accuracy): TopKClassificationAccuracy()\n (preprocessor): AudioToMelSpectrogramPreprocessor(\n (featurizer): FilterbankFeatures()\n )\n (encoder): ConvASREncoder(\n (encoder): Sequential(\n (0): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(80, 80, kernel_size=(3,), stride=(1,), padding=(1,), groups=80, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(80, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n (1): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(7,), stride=(1,), padding=(3,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (2): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(11,), stride=(1,), padding=(5,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (3): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): ReLU(inplace=True)\n (4): Dropout(p=0.1, inplace=False)\n (5): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (6): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (7): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (8): ReLU(inplace=True)\n (9): Dropout(p=0.1, inplace=False)\n (10): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(15,), stride=(1,), padding=(7,), groups=1024, bias=False)\n )\n (11): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (12): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (13): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=1024, out_features=128, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=128, out_features=1024, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (res): ModuleList(\n (0): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), bias=False)\n )\n (1): BatchNorm1d(1024, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.1, inplace=False)\n )\n )\n (4): JasperBlock(\n (mconv): ModuleList(\n (0): MaskedConv1d(\n (conv): Conv1d(1024, 1024, kernel_size=(1,), stride=(1,), groups=1024, bias=False)\n )\n (1): MaskedConv1d(\n (conv): Conv1d(1024, 3072, kernel_size=(1,), stride=(1,), bias=False)\n )\n (2): BatchNorm1d(3072, eps=0.001, momentum=0.1, affine=True, track_running_stats=True)\n (3): SqueezeExcite(\n (fc): Sequential(\n (0): Linear(in_features=3072, out_features=384, bias=False)\n (1): ReLU(inplace=True)\n (2): Linear(in_features=384, out_features=3072, bias=False)\n )\n (gap): AdaptiveAvgPool1d(output_size=1)\n )\n )\n (mout): Sequential(\n (0): ReLU(inplace=True)\n (1): Dropout(p=0.0, inplace=False)\n )\n )\n )\n )\n (decoder): SpeakerDecoder(\n (_pooling): AttentivePoolLayer(\n (attention_layer): Sequential(\n (0): TDNNModule(\n (conv_layer): Conv1d(9216, 128, kernel_size=(1,), stride=(1,))\n (activation): ReLU()\n (bn): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n )\n (1): Tanh()\n (2): Conv1d(128, 3072, kernel_size=(1,), stride=(1,))\n )\n )\n (emb_layers): ModuleList(\n (0): Sequential(\n (0): BatchNorm1d(6144, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n (1): Conv1d(6144, 192, kernel_size=(1,), stride=(1,))\n )\n )\n (final): Linear(in_features=192, out_features=16681, bias=False)\n )\n (_macro_accuracy): Accuracy()\n (spec_augmentation): SpectrogramAugmentation(\n (spec_augment): SpecAugment()\n )\n )\n )\n )\n)" 765 | }, 766 | "execution_count": 22, 767 | "metadata": {}, 768 | "output_type": "execute_result" 769 | } 770 | ], 771 | "source": [ 772 | "for model in all_models.keys():\n", 773 | " mdo = getattr(nemo_asr.models, model) \n", 774 | "nemo_asr.models.NeuralDiarizer.from_pretrained('diar_msdd_telephonic')" 775 | ], 776 | "metadata": { 777 | "collapsed": false, 778 | "ExecuteTime": { 779 | "end_time": "2024-10-31T19:46:00.085531700Z", 780 | "start_time": "2024-10-31T19:45:55.195086100Z" 781 | } 782 | }, 783 | "id": "31f417e55cfea96c" 784 | }, 785 | { 786 | "cell_type": "code", 787 | "execution_count": 14, 788 | "outputs": [ 789 | { 790 | "name": "stdout", 791 | "output_type": "stream", 792 | "text": [ 793 | "[NeMo I 2024-10-31 23:51:53 cloud:58] Found existing object C:\\Users\\Administrator\\.cache\\torch\\NeMo\\NeMo_1.21.0\\vad_multilingual_frame_marblenet\\797ff2f9c4ffbe205a462a226ed7c224\\vad_multilingual_frame_marblenet.nemo.\n", 794 | "[NeMo I 2024-10-31 23:51:53 cloud:64] Re-using file from: C:\\Users\\Administrator\\.cache\\torch\\NeMo\\NeMo_1.21.0\\vad_multilingual_frame_marblenet\\797ff2f9c4ffbe205a462a226ed7c224\\vad_multilingual_frame_marblenet.nemo\n", 795 | "[NeMo I 2024-10-31 23:51:53 common:913] Instantiating model from pre-trained checkpoint\n" 796 | ] 797 | }, 798 | { 799 | "name": "stderr", 800 | "output_type": "stream", 801 | "text": [ 802 | "[NeMo W 2024-10-31 23:51:53 modelPT:161] If you intend to do training or fine-tuning, please call the ModelPT.setup_training_data() method and provide a valid configuration file to setup the train data loader.\n", 803 | " Train config : \n", 804 | " manifest_filepath:\n", 805 | " - /manifests/vad/fisher_dur20_spk2_ovl0.1_sln0.4_seed0_500h_train.json\n", 806 | " - /manifests/vad/ls960_dur20_spk2_sln0.3x0.02_500h_train.json\n", 807 | " - /manifests/vad/french_train_40ms_cleaned.json\n", 808 | " - /manifests/vad/german_train_40ms.json\n", 809 | " - /manifests/vad/mandarin_train_40ms.json\n", 810 | " - /manifests/vad/russian_train_40ms.json\n", 811 | " - /manifests/vad/spanish_train_40ms.json\n", 812 | " sample_rate: 16000\n", 813 | " labels:\n", 814 | " - '0'\n", 815 | " - '1'\n", 816 | " batch_size: 64\n", 817 | " shuffle: true\n", 818 | " is_tarred: false\n", 819 | " tarred_audio_filepaths: na\n", 820 | " tarred_shard_strategy: scatter\n", 821 | " shuffle_n: 2048\n", 822 | " num_workers: 8\n", 823 | " pin_memory: true\n", 824 | " bucketing_strategy: ''\n", 825 | " bucketing_batch_size: ''\n", 826 | " bucketing_weights: null\n", 827 | " augmentor:\n", 828 | " white_noise:\n", 829 | " prob: 0.9\n", 830 | " min_level: -90\n", 831 | " max_level: -46\n", 832 | " gain:\n", 833 | " prob: 0.8\n", 834 | " min_gain_dbfs: -20\n", 835 | " max_gain_dbfs: 10\n", 836 | " noise:\n", 837 | " prob: 0.6\n", 838 | " manifest_path:\n", 839 | " - /manifests/vad/white_noise_1ch_37h_drc.json\n", 840 | " - /manifests/vad/freesound_nonspeech_train.json\n", 841 | " - /manifests/vad/musan_train.json\n", 842 | " min_snr_db: 0\n", 843 | " max_snr_db: 20\n", 844 | " max_gain_db: 300.0\n", 845 | " \n", 846 | "[NeMo W 2024-10-31 23:51:53 modelPT:168] If you intend to do validation, please call the ModelPT.setup_validation_data() or ModelPT.setup_multiple_validation_data() method and provide a valid configuration file to setup the validation data loader(s). \n", 847 | " Validation config : \n", 848 | " manifest_filepath:\n", 849 | " - /manifests/vad/fisher_dur180_spks6_turnP0.85_ovl0.05x0.005_sln0.3x0.005_seed432_snr0_sv1.0x0.2x2.0_dev_wn_25h_20s.json\n", 850 | " - /manifests/vad/fisher_dur20_spk2_ovl0.1_sln0.4_seed2_dev_noisy_snr0_100h.json\n", 851 | " - /manifests/vad/ls960_dur20_spk2_sln0.3x0.02_500h_dev.json\n", 852 | " - /manifests/vad/french_dev_40ms.json\n", 853 | " - /manifests/vad/german_dev_40ms.json\n", 854 | " - /manifests/vad/mandarin_dev_40ms.json\n", 855 | " - /manifests/vad/russian_dev_40ms.json\n", 856 | " - /manifests/vad/spanish_dev_40ms.json\n", 857 | " sample_rate: 16000\n", 858 | " labels:\n", 859 | " - '0'\n", 860 | " - '1'\n", 861 | " batch_size: 64\n", 862 | " shuffle: false\n", 863 | " num_workers: 8\n", 864 | " pin_memory: true\n", 865 | " val_loss_idx: 0\n", 866 | " \n", 867 | "[NeMo W 2024-10-31 23:51:53 modelPT:174] Please call the ModelPT.setup_test_data() or ModelPT.setup_multiple_test_data() method and provide a valid configuration file to setup the test data loader(s).\n", 868 | " Test config : \n", 869 | " manifest_filepath:\n", 870 | " - /manifests/vad/fisher_dur180_spks6_turnP0.85_ovl0.05x0.005_sln0.3x0.005_seed432_snr0_sv1.0x0.2x2.0_dev_wn_25h_20s.json\n", 871 | " - /manifests/vad/fisher_dur20_spk2_ovl0.1_sln0.4_seed2_dev_noisy_snr0_100h.json\n", 872 | " - /manifests/vad/ls960_dur20_spk2_sln0.3x0.02_500h_dev.json\n", 873 | " - /manifests/vad/french_dev_40ms.json\n", 874 | " - /manifests/vad/german_dev_40ms.json\n", 875 | " - /manifests/vad/mandarin_dev_40ms.json\n", 876 | " - /manifests/vad/russian_dev_40ms.json\n", 877 | " - /manifests/vad/spanish_dev_40ms.json\n", 878 | " sample_rate: 16000\n", 879 | " labels:\n", 880 | " - '0'\n", 881 | " - '1'\n", 882 | " batch_size: 64\n", 883 | " shuffle: false\n", 884 | " num_workers: 8\n", 885 | " pin_memory: true\n", 886 | " test_loss_idx: 0\n", 887 | " \n" 888 | ] 889 | }, 890 | { 891 | "name": "stdout", 892 | "output_type": "stream", 893 | "text": [ 894 | "[NeMo I 2024-10-31 23:51:53 features:289] PADDING: 2\n", 895 | "[NeMo I 2024-10-31 23:51:53 classification_models:888] Using cross-entropy with weights: [1.0, 1.0]\n", 896 | "[NeMo I 2024-10-31 23:51:53 cross_entropy:55] Weighted Cross Entropy loss with weight tensor([1., 1.])\n", 897 | "[NeMo I 2024-10-31 23:51:53 save_restore_connector:249] Model EncDecFrameClassificationModel was successfully restored from C:\\Users\\Administrator\\.cache\\torch\\NeMo\\NeMo_1.21.0\\vad_multilingual_frame_marblenet\\797ff2f9c4ffbe205a462a226ed7c224\\vad_multilingual_frame_marblenet.nemo.\n" 898 | ] 899 | } 900 | ], 901 | "source": [ 902 | "model = nemo_asr.models.EncDecFrameClassificationModel.from_pretrained('vad_multilingual_frame_marblenet')" 903 | ], 904 | "metadata": { 905 | "collapsed": false, 906 | "ExecuteTime": { 907 | "end_time": "2024-10-31T23:51:54.001002500Z", 908 | "start_time": "2024-10-31T23:51:53.719736Z" 909 | } 910 | }, 911 | "id": "1941295268fff9ca" 912 | }, 913 | { 914 | "cell_type": "code", 915 | "execution_count": 13, 916 | "outputs": [], 917 | "source": [ 918 | "model.save_to('hey.nemo')" 919 | ], 920 | "metadata": { 921 | "collapsed": false, 922 | "ExecuteTime": { 923 | "end_time": "2024-10-31T23:51:43.291943100Z", 924 | "start_time": "2024-10-31T23:51:43.228290600Z" 925 | } 926 | }, 927 | "id": "85f6618f9dc0d10a" 928 | }, 929 | { 930 | "cell_type": "code", 931 | "execution_count": null, 932 | "outputs": [], 933 | "source": [], 934 | "metadata": { 935 | "collapsed": false 936 | }, 937 | "id": "5f29f6137daa5ffb" 938 | } 939 | ], 940 | "metadata": { 941 | "kernelspec": { 942 | "display_name": "Python 3", 943 | "language": "python", 944 | "name": "python3" 945 | }, 946 | "language_info": { 947 | "codemirror_mode": { 948 | "name": "ipython", 949 | "version": 2 950 | }, 951 | "file_extension": ".py", 952 | "mimetype": "text/x-python", 953 | "name": "python", 954 | "nbconvert_exporter": "python", 955 | "pygments_lexer": "ipython2", 956 | "version": "2.7.6" 957 | } 958 | }, 959 | "nbformat": 4, 960 | "nbformat_minor": 5 961 | } 962 | --------------------------------------------------------------------------------