DrishtiSharma/wav2vec2-large-xls-r-300m-pa-IN-dx1

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - PA-IN dataset. It achieves the following results on the evaluation set:

• Loss: 1.0855
• Wer: 0.4755

Evaluation Commands

1. To evaluate on mozilla-foundation/common_voice_8_0 with test split

python eval.py --model_id DrishtiSharma/wav2vec2-large-xls-r-300m-pa-IN-dx1 --dataset mozilla-foundation/common_voice_8_0 --config pa-IN --split test --log_outputs

2. To evaluate on speech-recognition-community-v2/dev_data

Punjabi language isn't available in speech-recognition-community-v2/dev_data

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0003
• train_batch_size: 16
• 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: 1200
• num_epochs: 100.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.4607	9.26	500	2.7746	1.0416
0.3442	18.52	1000	0.9114	0.5911
0.2213	27.78	1500	0.9687	0.5751
0.1242	37.04	2000	1.0204	0.5461
0.0998	46.3	2500	1.0250	0.5233
0.0727	55.56	3000	1.1072	0.5382
0.0605	64.81	3500	1.0588	0.5073
0.0458	74.07	4000	1.0818	0.5069
0.0338	83.33	4500	1.0948	0.5108
0.0223	92.59	5000	1.0986	0.4775

Framework versions

• Transformers 4.17.0.dev0
• Pytorch 1.10.2+cu102
• Datasets 1.18.2.dev0
• Tokenizers 0.11.0