xls-r-300m-model-D

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

xls-r-300m-model-D 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

18% 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: 865be074

Model card

From the Hugging Face model README

xls-r-300m-model-D

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: 1.7217
Wer: 0.9969
Cer: 0.6902

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: 2e-05
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
32.0223	1.0	47	31.7006	1.0	2.6627
30.6191	2.0	94	28.7189	1.0	0.9802
23.48	3.0	141	14.8504	1.0	1.0
13.7384	4.0	188	8.6016	1.0	1.0
9.6022	5.0	235	6.3337	1.0	1.0
7.0718	6.0	282	5.0846	1.0	1.0
5.3429	7.0	329	4.2855	1.0	1.0
4.3829	8.0	376	3.8110	1.0	1.0
3.9108	9.0	423	3.5171	1.0	1.0
3.5356	10.0	470	3.3415	1.0	1.0
3.3748	11.0	517	3.2509	1.0	1.0
3.2788	12.0	564	3.2040	1.0	1.0
3.2254	13.0	611	3.1804	1.0	1.0
3.1871	14.0	658	3.1688	1.0	1.0
3.1704	15.0	705	3.1607	1.0	1.0
3.1741	16.0	752	3.1564	1.0	1.0
3.1516	17.0	799	3.1502	1.0	1.0
3.1504	18.0	846	3.1475	1.0	1.0
3.1475	19.0	893	3.1443	1.0	1.0
3.1404	20.0	940	3.1384	1.0	1.0
3.1413	21.0	987	3.1333	1.0	1.0
3.1307	22.0	1034	3.1230	1.0	1.0
3.1252	23.0	1081	3.0888	1.0	1.0
3.1063	24.0	1128	3.0454	1.0	1.0
3.0689	25.0	1175	2.9970	1.0	1.0
3.0212	26.0	1222	2.9566	1.0	1.0
2.9718	27.0	1269	2.8838	1.0	1.0
2.9241	28.0	1316	2.8342	1.0	1.0
2.874	29.0	1363	2.7721	1.0	0.9492
2.8058	30.0	1410	2.7171	1.0	0.9005
2.7639	31.0	1457	2.6737	1.0	0.8525
2.7273	32.0	1504	2.6319	1.0	0.8546
2.6737	33.0	1551	2.5578	1.0	0.8440
2.6197	34.0	1598	2.5156	1.0	0.8391
2.5877	35.0	1645	2.4513	1.0	0.8335
2.5275	36.0	1692	2.3993	1.0	0.8405
2.4871	37.0	1739	2.3387	1.0	0.8292
2.4367	38.0	1786	2.2844	1.0	0.8384
2.3918	39.0	1833	2.2465	1.0	0.8342
2.3633	40.0	1880	2.1934	1.0	0.8215
2.3201	41.0	1927	2.1387	1.0	0.8215
2.2691	42.0	1974	2.1210	1.0	0.8158
2.2387	43.0	2021	2.0740	1.0	0.8073
2.195	44.0	2068	2.0135	1.0	0.7939
2.1438	45.0	2115	1.9877	1.0	0.7855
2.1288	46.0	2162	1.9605	0.9969	0.7530
2.1215	47.0	2209	1.9264	0.9969	0.7481
2.0892	48.0	2256	1.8813	1.0	0.7318
2.0759	49.0	2303	1.8614	1.0	0.7255
2.0326	50.0	2350	1.8368	1.0	0.7198
2.0058	51.0	2397	1.8204	0.9969	0.7121
2.0022	52.0	2444	1.7939	1.0	0.7022
2.0073	53.0	2491	1.7808	0.9969	0.7036
1.9757	54.0	2538	1.7677	1.0	0.7001
1.9783	55.0	2585	1.7468	1.0	0.6958
1.9394	56.0	2632	1.7406	1.0	0.6958
1.9373	57.0	2679	1.7265	0.9969	0.6923
1.936	58.0	2726	1.7197	0.9969	0.6888
1.9291	59.0	2773	1.7206	0.9969	0.6916
1.9272	60.0	2820	1.7217	0.9969	0.6902

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