Edit model card

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

Visualize in Weights & Biases

MMS-Adapter-Testing

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: 0.9386
  • Wer: 0.6378

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
  • train_batch_size: 2
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 10
  • num_epochs: 1
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
5.768 0.0150 100 2.1552 0.9390
2.3489 0.0299 200 1.1714 0.7209
1.7924 0.0449 300 1.1720 0.7586
1.7483 0.0598 400 1.0868 0.7237
1.8404 0.0748 500 1.0824 0.6963
1.8122 0.0897 600 1.0771 0.6866
1.7504 0.1047 700 1.0705 0.6970
1.6675 0.1196 800 1.0688 0.6913
1.6123 0.1346 900 1.0446 0.6888
1.6237 0.1495 1000 1.0586 0.7034
1.6714 0.1645 1100 1.0562 0.6866
1.8129 0.1795 1200 1.0363 0.6891
1.7839 0.1944 1300 1.0374 0.6631
1.7305 0.2094 1400 1.0211 0.6834
1.5496 0.2243 1500 1.0225 0.6856
1.5106 0.2393 1600 1.0387 0.7127
1.7517 0.2542 1700 1.0561 0.6898
1.7117 0.2692 1800 1.0303 0.6866
1.6854 0.2841 1900 1.0240 0.6888
1.5186 0.2991 2000 1.0207 0.6873
1.5631 0.3140 2100 0.9964 0.6677
1.6909 0.3290 2200 1.0090 0.6738
1.5698 0.3440 2300 1.0016 0.6809
1.6702 0.3589 2400 0.9996 0.6749
1.628 0.3739 2500 1.0074 0.6699
1.8025 0.3888 2600 1.0312 0.6934
1.5986 0.4038 2700 0.9871 0.6667
1.5687 0.4187 2800 0.9893 0.6567
1.6444 0.4337 2900 0.9943 0.6674
1.5869 0.4486 3000 0.9831 0.6706
1.443 0.4636 3100 1.0192 0.7045
1.569 0.4785 3200 0.9783 0.6635
1.5302 0.4935 3300 0.9898 0.6727
1.5879 0.5084 3400 0.9773 0.6670
1.5739 0.5234 3500 0.9837 0.6895
1.5684 0.5384 3600 0.9836 0.6667
1.6397 0.5533 3700 0.9673 0.6578
1.5639 0.5683 3800 0.9888 0.6599
1.6773 0.5832 3900 0.9788 0.6613
1.5069 0.5982 4000 0.9801 0.6542
1.4801 0.6131 4100 0.9587 0.6545
1.7308 0.6281 4200 0.9599 0.6706
1.4852 0.6430 4300 0.9728 0.6663
1.4654 0.6580 4400 0.9468 0.6417
1.801 0.6729 4500 0.9591 0.6556
2.0928 0.6879 4600 0.9857 0.6670
1.561 0.7029 4700 0.9550 0.6503
1.6623 0.7178 4800 0.9587 0.6524
1.5252 0.7328 4900 0.9551 0.6531
1.5539 0.7477 5000 0.9660 0.6513
1.5571 0.7627 5100 0.9557 0.6531
1.6584 0.7776 5200 0.9649 0.6563
1.5072 0.7926 5300 0.9604 0.6481
1.5362 0.8075 5400 0.9457 0.6314
1.4772 0.8225 5500 0.9491 0.6449
1.3731 0.8374 5600 0.9609 0.6478
1.5795 0.8524 5700 0.9568 0.6567
1.4013 0.8674 5800 0.9457 0.6406
1.5817 0.8823 5900 0.9437 0.6513
1.4211 0.8973 6000 0.9433 0.6381
1.4341 0.9122 6100 0.9420 0.6353
1.4818 0.9272 6200 0.9407 0.6456
1.5241 0.9421 6300 0.9400 0.6381
1.575 0.9571 6400 0.9374 0.6392
1.5232 0.9720 6500 0.9385 0.6364
1.8634 0.9870 6600 0.9386 0.6378

Framework versions

  • Transformers 4.41.0.dev0
  • Pytorch 2.1.2
  • Datasets 2.18.0
  • Tokenizers 0.19.1
Downloads last month
0
Safetensors
Model size
965M params
Tensor type
F32
·
Inference API
or
This model can be loaded on Inference API (serverless).

Finetuned from