Modelpublic

nrshoudi/wav2vec2-large-xls-r-300m-Arabic-phoneme-based-MDD

sourceHugging Faceapache-2.0updated 3y agoView on Hugging Face

0likes50downloads

Model Card

wav2vec2-large-xls-r-300m-Arabic-phoneme-based-MDD

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.7800
—Per: 0.1135

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: 0.0005
—trainbatchsize: 8
—evalbatchsize: 6
—seed: 42
—gradientaccumulationsteps: 4
—totaltrainbatch_size: 32
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 250
—num_epochs: 40.0

Training results

Training Loss	Epoch	Step	Validation Loss	Per
4.5228	1.0	546	2.1723	0.6314
1.2389	2.0	1093	0.9571	0.2597
0.7931	3.0	1640	0.8440	0.2246
0.6438	4.0	2187	0.7831	0.2045
0.5584	5.0	2733	0.7660	0.1922
0.5062	6.0	3280	0.7193	0.1724
0.4596	7.0	3827	0.7373	0.1720
0.4227	8.0	4374	0.6829	0.1629
0.3832	9.0	4920	0.7181	0.1608
0.3617	10.0	5467	0.7043	0.1591
0.3495	11.0	6014	0.7295	0.1566
0.3282	12.0	6561	0.6897	0.1508
0.3086	13.0	7107	0.7353	0.1554
0.2911	14.0	7654	0.7144	0.1477
0.2801	15.0	8201	0.6988	0.1442
0.2658	16.0	8748	0.7061	0.1475
0.252	17.0	9294	0.7090	0.1403
0.2487	18.0	9841	0.7032	0.1363
0.2363	19.0	10388	0.7087	0.1395
0.222	20.0	10935	0.6982	0.1345
0.2152	21.0	11481	0.6964	0.1361
0.2063	22.0	12028	0.7246	0.1341
0.1958	23.0	12575	0.7331	0.1347
0.1866	24.0	13122	0.7493	0.1326
0.1786	25.0	13668	0.7536	0.1381
0.1751	26.0	14215	0.7345	0.1308
0.169	27.0	14762	0.7274	0.1251
0.1616	28.0	15309	0.7590	0.1293
0.1589	29.0	15855	0.7330	0.1243
0.1495	30.0	16402	0.7517	0.1228
0.1415	31.0	16949	0.7454	0.1208
0.1376	32.0	17496	0.7827	0.1254
0.1337	33.0	18042	0.7523	0.1221
0.128	34.0	18589	0.7752	0.1208
0.1262	35.0	19136	0.7716	0.1174
0.1196	36.0	19683	0.7620	0.1164
0.1161	37.0	20229	0.7792	0.1164
0.1117	38.0	20776	0.7800	0.1140
0.1103	39.0	21323	0.7716	0.1134
0.1074	39.95	21840	0.7800	0.1135

Framework versions

—Transformers 4.32.0
—Pytorch 2.0.1+cu118
—Datasets 1.18.3
—Tokenizers 0.13.3