xls-r-300m-model-3

dizzyme/xls-r-300m-model-3

xls-r-300m-model-3 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

28% 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: Jan 20, 2024
Updated: Jan 22, 2024
SHA: c98ab9a0

Model card

From the Hugging Face model README

xls-r-300m-model-3

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: 0.0588
Wer: 0.3824
Cer: 0.1108

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
27.5714	1.0	47	14.0582	1.0	1.0
9.1376	2.0	94	3.9617	1.0	1.0
3.5313	3.0	141	3.1685	1.0	1.0
3.1824	4.0	188	3.1454	1.0	1.0
3.1515	5.0	235	3.2191	1.0	1.0
3.1334	6.0	282	3.0060	1.0	1.0
2.9592	7.0	329	2.8529	1.0	1.0
2.8578	8.0	376	2.6646	1.0	1.0
2.7037	9.0	423	2.5452	1.0	0.9781
2.4728	10.0	470	2.2896	1.0	0.8737
2.0318	11.0	517	1.5268	1.0	0.6782
1.39	12.0	564	1.0841	1.0	0.4622
0.8994	13.0	611	0.5316	1.0	0.3691
0.6888	14.0	658	0.5907	0.9498	0.2576
0.4924	15.0	705	0.3629	0.9154	0.2548
0.4083	16.0	752	0.2853	0.7774	0.2421
0.3278	17.0	799	0.1644	0.7524	0.2174
0.2391	18.0	846	0.1845	0.6082	0.1771
0.2025	19.0	893	0.1324	0.5549	0.1334
0.2015	20.0	940	0.2035	0.5361	0.1658
0.1966	21.0	987	0.1196	0.5392	0.1313
0.2209	22.0	1034	0.1193	0.5078	0.1009
0.1588	23.0	1081	0.0958	0.5862	0.2040
0.1134	24.0	1128	0.1008	0.5455	0.1376
0.1106	25.0	1175	0.0783	0.4734	0.1503
0.1249	26.0	1222	0.0652	0.4138	0.0854
0.0797	27.0	1269	0.1045	0.4734	0.1284
0.1123	28.0	1316	0.1391	0.4389	0.1052
0.0916	29.0	1363	0.0699	0.4138	0.1023
0.0718	30.0	1410	0.0789	0.3762	0.0847
0.0927	31.0	1457	0.0839	0.4577	0.1179
0.1101	32.0	1504	0.1126	0.4138	0.1052
0.0718	33.0	1551	0.0711	0.4263	0.1066
0.067	34.0	1598	0.1456	0.4263	0.1200
0.0727	35.0	1645	0.1152	0.4138	0.1228
0.045	36.0	1692	0.1489	0.4295	0.1355
0.0607	37.0	1739	0.0833	0.3762	0.0812
0.0465	38.0	1786	0.1160	0.4075	0.1235
0.0564	39.0	1833	0.1005	0.4169	0.1277
0.05	40.0	1880	0.0842	0.4107	0.1454
0.033	41.0	1927	0.0838	0.3856	0.1164
0.0329	42.0	1974	0.0968	0.4075	0.1482
0.0353	43.0	2021	0.0992	0.3887	0.1023
0.0371	44.0	2068	0.1162	0.3981	0.1362
0.0285	45.0	2115	0.0999	0.3950	0.1362
0.0324	46.0	2162	0.0875	0.3918	0.1348
0.0305	47.0	2209	0.0789	0.4044	0.1560
0.0292	48.0	2256	0.0772	0.3981	0.1496
0.0225	49.0	2303	0.0767	0.3950	0.1553
0.0271	50.0	2350	0.0892	0.3950	0.1136
0.0231	51.0	2397	0.0905	0.3824	0.1023
0.0198	52.0	2444	0.0908	0.3793	0.1044
0.0263	53.0	2491	0.0750	0.3824	0.1023
0.022	54.0	2538	0.0666	0.3793	0.1016
0.0235	55.0	2585	0.0621	0.3762	0.0995
0.0267	56.0	2632	0.0563	0.3824	0.1073
0.0379	57.0	2679	0.0578	0.3856	0.1108
0.0243	58.0	2726	0.0596	0.3856	0.1115
0.0261	59.0	2773	0.0587	0.3824	0.1094
0.0249	60.0	2820	0.0588	0.3824	0.1108

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: Jan 20, 2024
Updated: Jan 22, 2024