metadata

license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
  - generated_from_trainer
datasets:
  - common_voice_17_0
metrics:
  - wer
model-index:
  - name: xls-r-300m-hbs-ar-unfrozen-batch16
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_17_0
          type: common_voice_17_0
          config: hsb
          split: test
          args: hsb
        metrics:
          - name: Wer
            type: wer
            value: 0.46954076850984067

xls-r-300m-hbs-ar-unfrozen-batch16

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.7763
Wer: 0.4695
Cer: 0.1093

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: 0.0003
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.3679	3.2258	100	3.2752	1.0	1.0
3.0469	6.4516	200	2.9638	1.0	0.9902
0.5512	9.6774	300	0.7542	0.7664	0.1947
0.3029	12.9032	400	0.6819	0.6432	0.1584
0.1903	16.1290	500	0.7312	0.6361	0.1572
0.1464	19.3548	600	0.7223	0.5916	0.1456
0.1205	22.5806	700	0.7566	0.5738	0.1416
0.091	25.8065	800	0.7472	0.5527	0.1308
0.0686	29.0323	900	0.7029	0.5452	0.1337
0.0598	32.2581	1000	0.7889	0.5464	0.1309
0.0607	35.4839	1100	0.8012	0.5672	0.1412
0.0557	38.7097	1200	0.7628	0.5302	0.1333
0.0421	41.9355	1300	0.7861	0.5258	0.1265
0.0532	45.1613	1400	0.7843	0.5314	0.1272
0.0298	48.3871	1500	0.7888	0.5279	0.1253
0.0543	51.6129	1600	0.7847	0.5295	0.1290
0.0404	54.8387	1700	0.7314	0.5246	0.1249
0.0522	58.0645	1800	0.7505	0.5134	0.1222
0.0275	61.2903	1900	0.7588	0.5082	0.1202
0.0786	64.5161	2000	0.7733	0.4930	0.1171
0.0439	67.7419	2100	0.7953	0.4977	0.1133
0.0418	70.9677	2200	0.7664	0.4897	0.1126
0.0399	74.1935	2300	0.7599	0.4845	0.1100
0.0211	77.4194	2400	0.7747	0.4763	0.1115
0.0225	80.6452	2500	0.7607	0.4702	0.1094
0.0446	83.8710	2600	0.7583	0.4768	0.1103
0.0236	87.0968	2700	0.7824	0.4754	0.1102
0.0267	90.3226	2800	0.7861	0.4726	0.1110
0.0255	93.5484	2900	0.7928	0.4712	0.1106
0.0254	96.7742	3000	0.7834	0.4684	0.1102
0.0137	100.0	3100	0.7763	0.4695	0.1093

Framework versions

Transformers 4.42.0.dev0
Pytorch 2.3.1+cu121
Datasets 2.19.2
Tokenizers 0.19.1