CoolFace
Modelpublic

RuudVelo/wav2vec2-large-xls-r-1b-nl

sourceHugging Faceapache-2.0updated 5y agoView on Hugging Face
0likes72downloads
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. -->

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the MOZILLA-FOUNDATION/COMMONVOICE8_0 - NL dataset. This model is also available with a language model which improves these results. This model can be found at https://huggingface.co/RuudVelo/wav2vec2-large-xls-r-1b-nl-lm. The Common Voice 8 Dutch test Wer is 9.73 of that model. It achieves the following results on the evaluation set:

  • —Loss: 0.1479
  • —Wer: 0.1156

Model description

Model fine-tuned using the wav2vec-als-r-1b model architecture

Intended uses & limitations

More information needed

Training and evaluation data

Model has been trained on Common Voice 8 Dutch

Training procedure

Training hyperparameters

Model parameters can be found under Files and versions in the run.sh file.

Training results

Training LossEpochStepValidation LossWer
1.22230.525000.38660.3425
1.07481.0310000.25740.2169
1.04161.5515000.21770.1946
0.99512.0620000.20080.1760
0.9752.5825000.19610.1751
0.94613.130000.19890.1782
0.93813.6135000.19280.1699
0.9344.1340000.19230.1633
0.93224.6445000.18710.1634
0.90125.1650000.18900.1702
0.90455.6855000.18820.1740
0.88266.1960000.18560.1575
0.88486.7165000.18610.1617
0.87237.2270000.19270.1646
0.87257.7475000.17980.1531
0.85738.2680000.17810.1587
0.86338.7785000.18520.1628
0.86039.2990000.18330.1601
0.84219.895000.17880.1543
0.840410.32100000.18440.1556
0.834210.84105000.17700.1538
0.816111.35110000.18210.1567
0.837111.87115000.19090.1629
0.808312.38120000.17780.1498
0.80612.9125000.18020.1547
0.801313.42130000.18590.1584
0.791313.93135000.18750.1517
0.806314.45140000.17990.1571
0.799114.96145000.17920.1538
0.784315.48150000.17530.1464
0.790516.0155000.17840.1508
0.780816.51160000.17710.1485
0.774317.03165000.17950.1491
0.783317.54170000.17220.1484
0.776318.06175000.17670.1518
0.769818.58180000.17200.1460
0.757119.09185000.17350.1478
0.767319.61190000.18170.1511
0.741520.12195000.17630.1481
0.75120.64200000.17420.1484
0.756321.16205000.18100.1611
0.742321.67210000.18170.1557
0.724222.19215000.16900.1446
0.725122.7220000.16840.1446
0.730223.22225000.17350.1430
0.73323.74230000.17200.1454
0.712824.25235000.16680.1383
0.718424.77240000.16350.1377
0.701525.28245000.16460.1389
0.719825.8250000.17750.1462
0.717826.32255000.17050.1419
0.719926.83260000.16490.1416
0.698127.35265000.17240.1418
0.688627.86270000.16330.1382
0.692228.38275000.16980.1420
0.683328.9280000.16110.1351
0.679829.41285000.16390.1365
0.671129.93290000.16680.1358
0.676230.44295000.16820.1355
0.659430.96300000.16290.1345
0.666431.48305000.16250.1321
0.683831.99310000.15970.1372
0.660332.51315000.15830.1302
0.646833.02320000.15950.1322
0.646433.54325000.16090.1315
0.662334.06330000.16220.1366
0.641434.57335000.15870.1330
0.624235.09340000.16140.1337
0.63235.6345000.15680.1272
0.634636.12350000.15830.1274
0.614336.64355000.15760.1264
0.620837.15360000.16210.1263
0.618537.67365000.16230.1270
0.612838.18370000.16040.1268
0.615138.7375000.15930.1246
0.608239.22380000.15320.1238
0.639.73385000.15240.1224
0.603240.25390000.15210.1212
0.601640.76395000.15510.1215
0.600941.28400000.15230.1215
0.587541.8405000.15410.1216
0.60842.31410000.15360.1209
0.587642.83415000.15670.1211
0.571443.34420000.15320.1217
0.575643.86425000.15160.1196
0.571944.38430000.14910.1191
0.582944.89435000.14970.1193
0.566445.41440000.14870.1173
0.570745.92445000.14700.1164
0.569646.44450000.14790.1161
0.576746.96455000.14920.1175
0.557347.47460000.14710.1165
0.562547.99465000.14840.1168
0.567148.5470000.14740.1162
0.548449.02475000.14790.1158
0.55549.54480000.14770.1157

Framework versions

  • —Transformers 4.17.0.dev0
  • —Pytorch 1.10.2+cu102
  • —Datasets 1.18.3
  • —Tokenizers 0.11.0