CoolFace
Modelpublic

atariq701/csalt-voice-noLID

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
0likes3downloads
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. -->

csalt-voice-noLID

This model is a fine-tuned version of openai/whisper-medium on an unknown dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.4769
  • —Wer: 17.0668

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: 1e-05
  • —trainbatchsize: 2
  • —evalbatchsize: 8
  • —seed: 42
  • —gradientaccumulationsteps: 8
  • —totaltrainbatch_size: 16
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 500
  • —num_epochs: 100
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
2.48280.9811261.336945.4250
1.18132.0530.897146.8527
0.85932.9811790.702763.2057
0.6184.01060.570137.6379
0.43894.98111320.428821.4796
0.15326.01590.301020.8955
0.08386.98111850.300519.0136
0.04898.02120.302722.6476
0.03298.98112380.312821.6742
0.023810.02650.322818.6243
0.01610.98112910.323518.5594
0.013312.03180.314518.2349
0.011612.98113440.339416.7424
0.011714.03710.341619.1434
0.01114.98113970.372818.8838
0.008916.04240.350818.2349
0.010316.98114500.369820.3115
0.012218.04770.368620.1168
0.014618.98115030.373519.1434
0.015420.05300.383019.5328
0.011520.98115560.380920.3764
0.008222.05830.398219.9870
0.006622.98116090.393619.0785
0.004824.06360.401819.8572
0.005524.98116620.382918.1051
0.00526.06890.372117.4562
0.004226.98117150.375917.9104
0.003528.07420.393017.7807
0.002428.98117680.398718.2349
0.002430.07950.415717.2615
0.001430.98118210.411417.0019
0.001232.08480.412316.8722
0.000932.98118740.421017.5211
0.000934.09010.418217.3264
0.000834.98119270.417617.3913
0.000836.09540.416817.4562
0.000436.98119800.422217.3264
0.000438.010070.425217.5860
0.000338.981110330.427617.2615
0.000340.010600.429117.5211
0.000340.981110860.429817.3913
0.000342.011130.430817.3913
0.000342.981111390.432517.0668
0.000344.011660.433717.0668
0.000344.981111920.434817.0668
0.000246.012190.435817.0668
0.000246.981112450.436417.0668
0.000248.012720.437817.0668
0.000248.981112980.438817.0668
0.000250.013250.440017.0019
0.000250.981113510.441117.0019
0.000252.013780.442117.0019
0.000252.981114040.442517.0019
0.000254.014310.443817.0668
0.000254.981114570.444617.0668
0.000256.014840.446117.0668
0.000256.981115100.446717.1317
0.000258.015370.447917.1317
0.000258.981115630.448817.1317
0.000260.015900.449717.1317
0.000260.981116160.450217.0019
0.000262.016430.451216.8722
0.000262.981116690.452017.0019
0.000264.016960.452816.8722
0.000264.981117220.454116.8722
0.000266.017490.454817.0668
0.000266.981117750.455317.0668
0.000268.018020.456017.1317
0.000268.981118280.456617.2615
0.000270.018550.457917.3913
0.000270.981118810.458217.3913
0.000272.019080.459017.3913
0.000272.981119340.459917.3913
0.000274.019610.460517.3264
0.000274.981119870.461217.3264
0.000276.020140.462017.3264
0.000176.981120400.468417.2615
0.000178.020670.471517.2615
0.000178.981120930.472617.0668
0.000180.021200.473117.0668
0.000180.981121460.473317.0668
0.000182.021730.473817.0668
0.000182.981121990.474117.0668
0.000184.022260.474417.0668
0.000184.981122520.474817.0668
0.000186.022790.475117.0668
0.000186.981123050.475417.0668
0.000188.023320.475617.0668
0.000188.981123580.475917.0668
0.000190.023850.476217.0668
0.000190.981124110.476217.0668
0.000192.024380.476517.0668
0.000192.981124640.476717.1317
0.000194.024910.476717.0668
0.000194.981125170.476917.0668
0.000196.025440.476917.1317
0.000196.981125700.476917.1317
0.000198.025970.476917.1317
0.000198.113226000.476917.0668

Framework versions

  • —Transformers 4.44.2
  • —Pytorch 2.4.1+cu118
  • —Datasets 3.1.0
  • —Tokenizers 0.19.1