Ir para o conteúdo
ModelListadosExterno
Baixar

Xlsr Luganda Waxal

This model is a fine-tuned version of sulaimank/wav2vec2-xlsr-swahili-400hr on an unknown dataset. It achieves the following results on the evaluation set: Loss: 0.1404 Wer: 0.2144 Cer: 0.0446 Zindi: 0.8705

Fontehuggingface.co/VinyVan/xlsr-luganda-waxal/tree/0cd0152e6a4cd0b1ace9432ec41b20ea819318ce
TipoOtherModelo basesulaimank/wav2vec2-xlsr-swahili-400hrVersãoRevision 0cd0152Cnota

Fonte do modelo

Tipo
Other
Tamanho do arquivo
1,18 GB
Modelo base
sulaimank/wav2vec2-xlsr-swahili-400hr
Versão
vRevision 0cd0152
Atualizado
27 de set. de 2026
Downloads
13
Fonte
Hugging Face

Descrição da fonte

xlsr-luganda-waxal

This model is a fine-tuned version of sulaimank/wav2vec2-xlsr-swahili-400hr on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1404
  • Wer: 0.2144
  • Cer: 0.0446
  • Zindi: 0.8705

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: 0.0001
  • train_batch_size: 32
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 64
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 100.0

Training results

Training LossEpochStepValidation LossWerCerZindi
4.75860.11165001.75331.00.79360.1032
1.15270.223310000.49040.73000.15410.5580
1.00810.334915000.42320.64160.13470.6119
0.88070.446620000.39160.59540.12500.6398
0.86720.558225000.36240.55550.11600.6643
0.80810.669930000.34480.54030.11040.6746
0.77360.781535000.33180.52130.10620.6862
0.75320.893240000.32070.50100.10340.6978
0.72971.004745000.31560.48120.10010.7093
0.69811.116350000.30630.47090.09830.7154
0.67851.228055000.29240.44810.09250.7297
0.64161.339660000.28590.44270.09200.7327
0.62931.451365000.28060.42970.08910.7406
0.62731.562970000.27140.41660.08640.7485
0.60361.674675000.26700.41240.08480.7514
0.60411.786280000.25900.39970.08250.7589
0.59801.897885000.26110.40300.08350.7567
0.56232.009490000.25730.38790.08110.7655
0.56172.121095000.25160.38590.08050.7668
0.56632.2327100000.24640.37000.07840.7758
0.54332.3443105000.24620.36710.07820.7774
0.52662.4560110000.23810.36460.07580.7798
0.52492.5676115000.23440.35400.07400.7860
0.52592.6792120000.23480.35080.07400.7876
0.51672.7909125000.23090.34810.07380.7891
0.50502.9025130000.22950.34460.07230.7915
0.48663.0141135000.22690.33750.07120.7957
0.47133.1257140000.22580.33830.07090.7954
0.47473.2374145000.22180.33170.06980.7992
0.47253.3490150000.22050.33050.06940.8001
0.49513.4606155000.21860.32400.06870.8036
0.45293.5723160000.21400.32080.06740.8059
0.46683.6839165000.21310.32040.06790.8059
0.46483.7956170000.20800.31200.06570.8112
0.46363.9072175000.20730.31090.06530.8119
0.44204.0188180000.20850.31100.06600.8115
0.44884.1304185000.20410.30870.06460.8134
0.42104.2420190000.20500.30590.06430.8149
0.43374.3537195000.20310.30420.06420.8158
0.45624.4653200000.20160.30200.06400.8170
0.43834.5770205000.19900.29740.06310.8198
0.43104.6886210000.19970.30150.06350.8175
0.43204.8003215000.19540.29650.06190.8208
0.41974.9119220000.19540.29200.06190.8230
0.41465.0234225000.19240.28930.06060.8251
0.40955.1351230000.19170.29090.06080.8242
0.39895.2467235000.19220.28580.06060.8268
0.40335.3584240000.19290.28720.06120.8258
0.39205.4700245000.19070.28720.06040.8262
0.39345.5817250000.18840.27950.05940.8305
0.38815.6933255000.18560.27950.05890.8308
0.38355.8050260000.18380.27540.05790.8333
0.38275.9166265000.18460.27620.05870.8326
0.37746.0281270000.18230.27460.05770.8338
0.37756.1398275000.18010.27350.05740.8345
0.37626.2514280000.18160.27270.05790.8347
0.37556.3631285000.17960.27030.05690.8364
0.36766.4747290000.18050.26910.05720.8368
0.38306.5864295000.17810.26580.05670.8387
0.37146.6980300000.17750.26820.05660.8376
0.36486.8096305000.17660.26530.05590.8394
0.36656.9213310000.17360.26210.05540.8413
0.36677.0328315000.17580.26340.05620.8402
0.36347.1445320000.17780.26310.05680.8400
0.35197.2561325000.17310.25740.05470.8440
0.36437.3678330000.17090.26230.05490.8414
0.34517.4794335000.17160.25710.05470.8441
0.34977.5910340000.17100.25460.05450.8454
0.35637.7027345000.17190.25670.05520.8440
0.33937.8143350000.16960.25510.05440.8453
0.34947.9260355000.16640.25550.05380.8454
0.34358.0375360000.16720.250.05320.8484
0.34158.1492365000.16620.25030.05330.8482
0.32978.2608370000.16550.25070.05260.8484
0.33168.3724375000.16580.25290.05320.8469
0.32918.4841380000.16590.24680.05250.8504
0.33328.5957385000.16430.24780.05230.8500
0.34808.7074390000.16220.24510.05200.8514
0.32658.8190395000.16230.24640.05210.8508
0.33238.9307400000.16250.24830.05220.8498
0.33119.0422405000.16170.24840.05200.8498
0.33429.1538410000.16250.24120.05130.8538
0.32489.2655415000.16130.24540.05210.8513
0.30819.3771420000.16210.24200.05130.8534
0.31039.4888425000.16030.23950.05050.8550
0.30889.6004430000.15900.24050.05100.8542
0.31009.7121435000.15900.24040.05040.8546
0.30359.8237440000.16090.24330.05130.8527
0.32309.9354445000.15970.23970.05090.8547
0.307010.0469450000.15920.24070.05060.8544
0.309810.1585455000.15780.24100.05090.8541
0.293210.2702460000.15710.24010.05000.8549
0.296110.3818465000.15620.23660.05000.8567
0.299810.4935470000.15510.23470.04960.8579
0.308810.6051475000.15490.23750.05020.8562
0.301410.7168480000.15490.23610.05000.8569
0.299610.8284485000.15490.23300.04930.8589
0.307110.9400490000.15490.23310.04950.8587
0.297611.0516495000.15500.23310.04940.8587
0.296511.1632500000.15460.23190.04950.8593
0.294311.2749505000.15330.23370.04870.8588
0.286211.3865510000.15270.22600.04780.8631
0.300411.4982515000.15290.22810.04860.8616
0.283111.6098520000.15180.23130.04870.8600
0.285711.7214525000.14940.22350.04770.8644
0.275111.8331530000.14950.22410.04750.8642
0.279911.9447535000.15030.22630.04780.8629
0.284312.0563540000.15000.22750.04750.8625
0.267912.1679545000.15020.23050.04820.8606
0.286112.2796550000.15000.22810.04790.8620
0.288212.3912555000.14920.22740.04780.8624
0.278412.5028560000.14950.22520.04790.8634
0.277012.6145565000.14830.22330.04730.8647
0.273312.7261570000.14940.22750.04780.8623
0.279612.8378575000.14800.22690.04790.8626
0.277912.9494580000.14730.22560.04720.8636
0.261213.0610585000.14880.22470.04770.8638
0.269013.1726590000.14870.22510.04760.8636
0.279913.2842595000.14790.22380.04710.8645
0.260713.3959600000.14810.22410.04710.8644
0.274913.5075605000.14860.22460.04750.8639
0.253813.6192610000.14730.22140.04670.8659
0.283213.7308615000.14580.22000.04680.8666
0.264013.8425620000.14470.21910.04600.8674
0.275413.9541625000.14670.22020.04650.8666
0.263614.0656630000.14580.22060.04680.8663
0.250914.1773635000.14600.22330.04690.8649
0.257714.2889640000.14410.21670.04610.8686
0.262414.4006645000.14470.21840.04590.8678
0.254214.5122...

Fontes

1 fonte
Hugging FacePrimárioVersão 0cd0152e6a4cd0b1ace9432ec41b20ea819318ce
huggingface.co/VinyVan/xlsr-luganda-waxal/tree/0cd0152e6a4cd0b1ace9432ec41b20ea819318ce

Verificado 30 de ago.

Artefatos de modelo

1 artefato

model.safetensors

safetensors · 1,18 GB · SHA-256 d1929a4c25fc…7075 · Hugging Face

Download file

Trechos de fonte

2 trechos
Readme30 de ago.huggingface.co/VinyVan/xlsr-luganda-waxal

--- library_name: transformers license: apache-2.0 base_model: sulaimank/wav2vec2-xlsr-swahili-400hr tags: - generated_from_trainer metrics: - wer model-index: - name: xlsr-luganda-waxal results: [] --- <!-- 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. --> # xlsr-luganda-waxal This model is a fine-tuned version of [sulaimank/wav2vec2-xlsr-swahili-400hr](https://huggingface.co/sulaimank/wav2vec2-xlsr-swahili-400hr) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 0.1404 - Wer: 0.2144 - Cer: 0.0446 - Zindi: 0.8705 ## 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: 0.0001 - train_batch_size: 32 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 2 - total_train_batch_size: 64 - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: linear - lr_scheduler_warmup_steps: 500 - num_epochs: 100.0 ### Training results | Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Zindi | |:-------------:|:-------:|:-----:|:---------------:|:------:|:------:|:------:| | 4.7586 | 0.1116 | 500 | 1.7533 | 1.0 | 0.7936 | 0.1032 | | 1.1527 | 0.2233 | 1000 | 0.4904 | 0.7300 | 0.1541 | 0.5580 | | 1.0081 | 0.3349 | 1500 | 0.4232 | 0.6416 | 0.1347 | 0.6119 | | 0.8807 | 0.4466 | 2000 | 0.3916 | 0.5954 | 0.1250 | 0.6398 | | 0.8672 | 0.5582 | 2500 | 0.3624 | 0.5555 | 0.1160 | 0.6643 | | 0.8081 | 0.6699 | 3000 | 0.3448 | 0.5403 | 0.1104 | 0.6746 | | 0.7736 | 0.7815 | 3500 | 0.3318 | 0.5213 | 0.1062 | 0.6862 | | 0.7532 | 0.8932 | 4000 | 0.3207 | 0.5010 | 0.1034 | 0.6978 | | 0.7297 | 1.0047 | 4500 | 0.3156 | 0.4812 | 0.1001 | 0.7093 | | 0.6981 | 1.1163 | 5000 | 0.3063 | 0.4709 | 0.0983 | 0.7154 | | 0.6785 | 1.2280 | 5500 | 0.2924 | 0.4481 | 0.0925 | 0.7297 | | 0.6416 | 1.3396 | 6000 | 0.2859 | 0.4427 | 0.0920 | 0.7327 | | 0.6293 | 1.4513 | 6500 | 0.2806 | 0.4297 | 0.0891 | 0.7406 | | 0.6273 | 1.5629 | 7000 | 0.2714 | 0.4166 | 0.0864 | 0.7485 | | 0.6036 |...

Note30 de ago.huggingface.co/VinyVan/xlsr-luganda-waxal

Source context: 13 downloads · 0 likes · Pipeline automatic-speech-recognition · Library transformers · Repo VinyVan/xlsr-luganda-waxal