CoolFace
Modelpublic

hajili/mms_1b_audio_books_aze

sourceHugging Facecc-by-nc-4.0updated 2y agoView on Hugging Face
0likes11downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

mms1baudiobooksaze

This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1030
  • Wer: 0.1235
  • Cer: 0.0240
  • Wer Book: 0.0991
  • Cer Book: 0.0165
  • Wer Cv: 0.2070
  • Cer Cv: 0.0468

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 0.001
  • trainbatchsize: 8
  • evalbatchsize: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lrschedulertype: linear
  • lrschedulerwarmup_steps: 20
  • num_epochs: 10
  • mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWerCerWer BookCer BookWer CvCer Cv
0.37020.251590.12050.14890.02800.13270.02160.20440.0474
0.3470.53180.11200.14070.02630.12140.01980.20700.0461
0.32590.754770.11810.14950.02750.12890.02080.21990.0479
0.31111.06360.11730.15450.02840.13570.02230.21860.0472
0.31521.257950.10850.13460.02490.11680.01800.19530.0463
0.29861.59540.10630.13960.02590.11980.01900.20700.0470
0.29971.7611130.11150.14280.02640.12060.01910.21860.0488
0.28672.0112720.11070.14450.02660.12100.01950.22510.0484
0.3032.2614310.11360.14340.02680.11950.01910.22510.0502
0.27362.5115900.11060.14160.02630.11830.01880.22120.0493
0.27772.7617490.10880.13780.02590.12140.01940.19400.0461
0.3273.0119080.11150.13660.02590.11300.01840.21730.0486
0.27893.2620670.10640.13630.02560.11760.01890.20050.0463
0.28283.5122260.10660.13550.02520.11570.01890.20310.0449
0.27683.7623850.10500.13020.02440.10850.01770.20440.0449
0.28584.0125440.10250.12900.02400.10660.01720.20570.0449
0.27854.2627030.10550.13340.02460.10960.01730.21470.0470
0.27334.5128620.10550.13340.02490.11380.01770.20050.0470
0.26244.7630210.10500.12900.02430.10780.01730.20180.0458
0.26765.0231800.10450.13050.02410.10550.01670.21600.0470
0.24575.2733390.10150.12870.02370.10740.01700.20180.0440
0.26375.5234980.10300.12900.02370.10550.01680.20960.0449
0.26465.7736570.10470.13460.02470.11150.01730.21350.0474
0.28316.0238160.10720.13020.02450.10850.01730.20440.0465
0.25026.2739750.10470.12670.02410.10430.01660.20310.0470
0.24276.5241340.10430.12990.02440.10660.01680.20960.0477
0.25766.7742930.10460.12490.02420.10210.01690.20310.0468
0.24667.0244520.10740.13080.02460.10430.01680.22120.0484
0.24637.2746110.10450.12550.02390.09830.01580.21860.0488
0.23777.5247700.10470.12960.02460.10250.01670.22250.0488
0.25057.7749290.10300.12790.02420.10320.01680.21220.0468
0.24798.0350880.10190.11910.02290.09380.01550.20570.0458
0.24518.2852470.10270.12400.02380.10060.01650.20440.0463
0.24428.5354060.10260.12610.02390.10090.01640.21220.0468
0.22268.7855650.10310.12460.02390.10130.01670.20440.0461
0.24629.0357240.10300.12520.02390.10060.01650.20960.0467
0.24399.2858830.10330.12430.02390.10090.01650.20440.0467
0.23639.5360420.10360.12200.02360.09600.01590.21090.0470
0.22219.7862010.10300.12350.02400.09910.01650.20700.0468

Framework versions

  • Transformers 4.40.0.dev0
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2