xls-r-300m-model-4

dizzyme/xls-r-300m-model-4

xls-r-300m-model-4 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

8% 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 22, 2024
Updated: Jan 22, 2024
SHA: 750ce513

Model card

From the Hugging Face model README

xls-r-300m-model-4

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: 0.1161
Wer: 0.2163
Cer: 0.0593

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
27.7684	1.0	47	14.3947	1.0	1.0
9.0101	2.0	94	4.2287	1.0	1.0
3.5959	3.0	141	3.1880	1.0	1.0
3.1801	4.0	188	3.1428	1.0	1.0
3.1508	5.0	235	3.0521	1.0	1.0
2.967	6.0	282	2.9580	1.0	1.0
2.7746	7.0	329	2.5700	1.0	1.0
2.449	8.0	376	2.0616	1.0	0.9358
1.6959	9.0	423	1.2861	1.0	0.6803
1.0742	10.0	470	0.7718	0.9937	0.4192
0.7025	11.0	517	0.4683	0.8025	0.2900
0.4039	12.0	564	0.3167	0.5298	0.2258
0.2477	13.0	611	0.2589	0.3793	0.1306
0.2605	14.0	658	0.2206	0.3918	0.1291
0.1525	15.0	705	0.2110	0.4013	0.1186
0.2108	16.0	752	0.1722	0.4107	0.1447
0.1358	17.0	799	0.1396	0.2978	0.0953
0.0716	18.0	846	0.1844	0.3511	0.0932
0.0567	19.0	893	0.1037	0.2602	0.0727
0.0675	20.0	940	0.0943	0.2382	0.0713
0.0651	21.0	987	0.1244	0.2539	0.0939
0.0893	22.0	1034	0.1289	0.2884	0.0967
0.0925	23.0	1081	0.1417	0.3229	0.1355
0.0659	24.0	1128	0.1326	0.3103	0.1313
0.0582	25.0	1175	0.0892	0.2821	0.1355
0.0526	26.0	1222	0.1754	0.2978	0.0797
0.0797	27.0	1269	0.1246	0.2602	0.0734
0.0366	28.0	1316	0.0984	0.2759	0.0875
0.0288	29.0	1363	0.0707	0.2257	0.0656
0.018	30.0	1410	0.0824	0.2257	0.0670
0.0329	31.0	1457	0.0924	0.2288	0.0748
0.0336	32.0	1504	0.1281	0.2539	0.0882
0.0299	33.0	1551	0.1219	0.2257	0.0593
0.0304	34.0	1598	0.1435	0.2351	0.0642
0.0732	35.0	1645	0.1605	0.2320	0.0734
0.0691	36.0	1692	0.0925	0.2132	0.0572
0.0284	37.0	1739	0.0951	0.2320	0.0692
0.0374	38.0	1786	0.0844	0.1881	0.0529
0.034	39.0	1833	0.0757	0.2351	0.0670
0.0172	40.0	1880	0.0632	0.2351	0.0642
0.0343	41.0	1927	0.1404	0.2288	0.0692
0.0424	42.0	1974	0.1433	0.2476	0.0762
0.0235	43.0	2021	0.1542	0.2571	0.1009
0.0219	44.0	2068	0.1100	0.2476	0.0819
0.0281	45.0	2115	0.1478	0.2320	0.0677
0.0205	46.0	2162	0.0797	0.2288	0.0706
0.0271	47.0	2209	0.1139	0.2382	0.0734
0.0288	48.0	2256	0.1264	0.2226	0.0670
0.0141	49.0	2303	0.1322	0.2288	0.0692
0.008	50.0	2350	0.1226	0.2226	0.0614
0.0098	51.0	2397	0.1146	0.1975	0.0621
0.006	52.0	2444	0.1116	0.2100	0.0635
0.0109	53.0	2491	0.1087	0.2163	0.0621
0.0058	54.0	2538	0.1176	0.2069	0.0565
0.0095	55.0	2585	0.1178	0.2132	0.0572
0.0061	56.0	2632	0.1193	0.2194	0.0579
0.011	57.0	2679	0.1152	0.2163	0.0579
0.0033	58.0	2726	0.1152	0.2163	0.0593
0.0045	59.0	2773	0.1160	0.2163	0.0586
0.0072	60.0	2820	0.1161	0.2163	0.0593

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 22, 2024
Updated: Jan 22, 2024