CoolFace
Modelpublic

chickenrice0721/whisper-large-v2-translate-zh-v0.1-lt

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
1likes8downloads
Model Card

whisper-large-v2-translate-zh-v0.1-lt-ct2

This model is a fine-tuned version of openai/whisper-large-v2.

Model description

3500小时 (日语音频,中文字幕) 数据微调, 翻译模式直出中文

Usage

task='translate', language='ja'

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • —learning_rate: 1e-05
  • —trainbatchsize: 32
  • —evalbatchsize: 32
  • —seed: 42
  • —gradientaccumulationsteps: 2
  • —totaltrainbatch_size: 64
  • —optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 4000
  • —dropout: 0.1
  • —masktimeprob: 0.05
  • —maskfeatureprob: 0.2
  • —conditiononprevioustextrate: 0.5

Training results

Training LossEpochStepValidation LossCerWer
1.7430.074010001.56310.82231.4517
1.60140.147920001.48080.67751.0950
1.55490.221930001.43810.67561.1158
1.52830.295840001.41740.69921.1137
1.4740.369850001.38490.65701.1369
1.41930.443760001.36570.65441.1339
1.41480.517770001.34770.63861.1647
1.37540.591680001.33920.62281.0461
1.34410.665690001.33620.61961.0609
1.35450.7395100001.31760.63541.2138
1.34980.8135110001.32360.66311.2232
1.310.8874120001.30200.61991.0018
1.32130.9614130001.29660.59221.0021
1.23751.0353140001.29000.60971.0639
1.23341.1093150001.29630.61501.0920
1.22771.1832160001.28880.60771.0929
1.20871.2572170001.27790.59541.0012
1.21311.3311180001.27220.57761.0075
1.20121.4051190001.27160.57261.0211
1.19121.4790200001.27070.60071.1538
1.21271.5530210001.27490.60861.0742
1.17891.6269220001.27970.57651.0072
1.15271.7009230001.27610.58551.0588
1.16931.7748240001.27010.56350.9928
1.17091.8488250001.26620.59801.0697
1.16371.9227260001.27490.58721.0392
1.15621.9967270001.25870.56511.0121
1.09292.0706280001.26680.58571.0139
1.12322.1446290001.27100.57420.9997
1.10452.2185300001.26560.56430.9897
1.08412.2925310001.26950.58351.0181
1.08682.3664320001.27070.56730.9964
1.09382.4404330001.26440.57120.9928
1.09382.5143340001.26620.57501.0109
1.08482.5883350001.26770.58411.0832
1.09142.6622360001.26380.58011.0299
1.06882.7362370001.25870.56941.0072
1.08562.8101380001.25810.56461.0057
1.10372.8841390001.25570.57711.0262
1.06522.9580400001.25660.56340.9979

Framework versions

  • —Transformers 4.41.1
  • —Pytorch 2.3.1+cu121
  • —Datasets 2.19.1
  • —Tokenizers 0.19.1