This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - JA dataset.
Kanji are converted into Hiragana using the pykakasi library during training and evaluation. The model can output both Hiragana and Katakana characters. Since there is no spacing, WER is not a suitable metric for evaluating performance and CER is more suitable.
On mozilla-foundation/common_voice_8_0 it achieved:
- cer: 23.64%
On speech-recognition-community-v2/dev_data it achieved:
- cer: 30.99%
It achieves the following results on the evaluation set:
- Loss: 0.5212
- Wer: 1.3068
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: 7.5e-05
- train_batch_size: 48
- 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: 2000
- num_epochs: 50.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
4.0974 | 4.72 | 1000 | 4.0178 | 1.9535 |
2.1276 | 9.43 | 2000 | 0.9301 | 1.2128 |
1.7622 | 14.15 | 3000 | 0.7103 | 1.5527 |
1.6397 | 18.87 | 4000 | 0.6729 | 1.4269 |
1.5468 | 23.58 | 5000 | 0.6087 | 1.2497 |
1.4885 | 28.3 | 6000 | 0.5786 | 1.3222 |
1.451 | 33.02 | 7000 | 0.5726 | 1.3768 |
1.3912 | 37.74 | 8000 | 0.5518 | 1.2497 |
1.3617 | 42.45 | 9000 | 0.5352 | 1.2694 |
1.3113 | 47.17 | 10000 | 0.5228 | 1.2781 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0
Evaluation Commands
- To evaluate on
mozilla-foundation/common_voice_8_0
with splittest
python ./eval.py --model_id AndrewMcDowell/wav2vec2-xls-r-300m-japanese --dataset mozilla-foundation/common_voice_8_0 --config ja --split test --log_outputs
- To evaluate on
mozilla-foundation/common_voice_8_0
with splittest
python ./eval.py --model_id AndrewMcDowell/wav2vec2-xls-r-300m-japanese --dataset speech-recognition-community-v2/dev_data --config de --split validation --chunk_length_s 5.0 --stride_length_s 1.0
- Downloads last month
- 21
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Dataset used to train AndrewMcDowell/wav2vec2-xls-r-300m-japanese
Evaluation results
- Test WER on Common Voice 8self-reported95.820
- Test CER on Common Voice 8self-reported23.640
- Test WER on Robust Speech Event - Dev Dataself-reported100.000
- Test CER on Robust Speech Event - Dev Dataself-reported30.990
- Test CER on Robust Speech Event - Dev Dataself-reported30.370
- Test CER on Robust Speech Event - Test Dataself-reported34.420