xls-r-300m-model-B

dizzyme/xls-r-300m-model-B

xls-r-300m-model-B is a automatic speech recognition model from dizzyme. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

17% of all-time downloads

All-time downloads

Public

Repo size

8.8 GB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.bin1.3 GB · 100%

At a glance

Task: Automatic Speech Recognition
Library: transformers
License: apache-2.0
Model type: wav2vec2
Access: Public
Created: Feb 1, 2024
Updated: Feb 1, 2024
SHA: 24761b81

Model card

From the Hugging Face model README

xls-r-300m-model-Bv1

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

Loss: 0.0132
Wer: 0.2884
Cer: 0.1969

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: 8
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_ratio: 0.1
num_epochs: 60
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
28.9378	1.0	47	14.4596	1.0	1.0
9.6471	2.0	94	4.1815	1.0	1.0
3.7029	3.0	141	3.1653	1.0	1.0
3.1683	4.0	188	3.1477	1.0	1.0
3.1645	5.0	235	3.1494	1.0	1.0
3.1589	6.0	282	3.1788	1.0	1.0
3.1643	7.0	329	3.0524	1.0	1.0
3.0349	8.0	376	2.8659	1.0	1.0
2.8767	9.0	423	2.9799	1.0	1.0
2.8025	10.0	470	2.6823	1.0	1.0
2.6742	11.0	517	2.5606	1.0	0.9704
2.4715	12.0	564	2.2902	1.0	0.8567
2.175	13.0	611	1.7539	1.0	0.7671
1.7504	14.0	658	1.4196	1.0	0.6683
1.4035	15.0	705	1.0294	1.0	0.5046
1.09	16.0	752	0.7075	0.9749	0.5095
0.9209	17.0	799	0.5899	0.9342	0.4785
0.726	18.0	846	0.4222	0.8339	0.3253
0.6051	19.0	893	0.3828	0.7461	0.3063
0.518	20.0	940	0.2521	0.6426	0.3529
0.4297	21.0	987	0.1871	0.5737	0.2519
0.3886	22.0	1034	0.2171	0.5329	0.2392
0.3754	23.0	1081	0.1561	0.4514	0.2322
0.3266	24.0	1128	0.0929	0.3699	0.2145
0.2661	25.0	1175	0.1174	0.3918	0.1842
0.2681	26.0	1222	0.0685	0.3699	0.2308
0.2139	27.0	1269	0.0848	0.3386	0.1849
0.2194	28.0	1316	0.1087	0.3793	0.2110
0.1974	29.0	1363	0.0671	0.3480	0.2167
0.1991	30.0	1410	0.0719	0.3511	0.2251
0.1656	31.0	1457	0.0459	0.2821	0.1235
0.1384	32.0	1504	0.0556	0.1818	0.0409
0.1505	33.0	1551	0.0378	0.2132	0.0607
0.1158	34.0	1598	0.0336	0.2696	0.1164
0.1366	35.0	1645	0.0483	0.2884	0.1426
0.1054	36.0	1692	0.0488	0.3292	0.1708
0.087	37.0	1739	0.0276	0.3292	0.1849
0.1021	38.0	1786	0.0242	0.3041	0.1821
0.1154	39.0	1833	0.0270	0.3197	0.1934
0.0984	40.0	1880	0.0188	0.3072	0.1941
0.1029	41.0	1927	0.0172	0.2947	0.2075
0.0658	42.0	1974	0.0257	0.2884	0.1969
0.0817	43.0	2021	0.0247	0.2947	0.1905
0.0793	44.0	2068	0.0330	0.2978	0.1934
0.0692	45.0	2115	0.0288	0.3009	0.2047
0.0719	46.0	2162	0.0138	0.2947	0.2018
0.0662	47.0	2209	0.0352	0.2790	0.1694
0.0824	48.0	2256	0.0180	0.2884	0.1800
0.062	49.0	2303	0.0123	0.2790	0.1715
0.0442	50.0	2350	0.0229	0.2853	0.1743
0.0588	51.0	2397	0.0213	0.3072	0.1976
0.055	52.0	2444	0.0185	0.2947	0.1941
0.0536	53.0	2491	0.0250	0.3009	0.1912
0.061	54.0	2538	0.0190	0.2947	0.1898
0.0483	55.0	2585	0.0132	0.2884	0.1955
0.0465	56.0	2632	0.0129	0.2884	0.2004
0.0411	57.0	2679	0.0165	0.2821	0.2018
0.0372	58.0	2726	0.0135	0.2884	0.1962
0.0365	59.0	2773	0.0133	0.2884	0.1969
0.0319	60.0	2820	0.0132	0.2884	0.1969

Framework versions

Transformers 4.28.0
Pytorch 2.1.0+cu121
Datasets 2.16.1
Tokenizers 0.13.3

Task: Automatic Speech Recognition
Library: transformers
Type: wav2vec2
License: apache-2.0
Created: Feb 1, 2024
Updated: Feb 1, 2024