wav2vec2-large-xlsr-coraa-exp-16

This model is a fine-tuned version of Edresson/wav2vec2-large-xlsr-coraa-portuguese on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.0160
• Wer: 1.0
• Cer: 0.9619
• Per: 1.0

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 150
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Per
38.847	1.0	14	48.0649	1.0002	3.1169	1.0002
38.847	2.0	28	47.9396	1.0002	3.1191	1.0002
38.847	3.0	42	47.7446	1.0004	3.0938	1.0004
38.847	4.0	56	47.3282	1.0006	3.0227	1.0006
38.847	5.0	70	46.4748	1.0004	2.4135	1.0004
38.847	6.0	84	44.8749	1.0	1.4413	1.0
38.847	7.0	98	42.6022	1.0	0.9460	1.0
37.4633	8.0	112	39.6558	1.0	0.8978	1.0
37.4633	9.0	126	35.7305	1.0	0.9338	1.0
37.4633	10.0	140	30.8796	1.0	0.9501	1.0
37.4633	11.0	154	26.3341	1.0	0.9510	1.0
37.4633	12.0	168	23.4116	1.0	0.9510	1.0
37.4633	13.0	182	21.1265	1.0	0.9510	1.0
37.4633	14.0	196	19.4607	1.0	0.9510	1.0
24.5313	15.0	210	18.0357	1.0	0.9510	1.0
24.5313	16.0	224	16.9469	1.0	0.9510	1.0
24.5313	17.0	238	16.1513	1.0	0.9510	1.0
24.5313	18.0	252	15.5951	1.0	0.9510	1.0
24.5313	19.0	266	15.0803	1.0	0.9510	1.0
24.5313	20.0	280	14.8270	1.0	0.9510	1.0
24.5313	21.0	294	14.4665	1.0	0.9510	1.0
14.1395	22.0	308	14.3680	1.0	0.9510	1.0
14.1395	23.0	322	14.1886	1.0	0.9510	1.0
14.1395	24.0	336	14.0514	1.0	0.9510	1.0
14.1395	25.0	350	14.0722	1.0	0.9510	1.0
14.1395	26.0	364	13.8134	1.0	0.9511	1.0
14.1395	27.0	378	13.6935	1.0	0.9522	1.0
14.1395	28.0	392	13.3832	1.0	0.9594	1.0
12.0999	29.0	406	12.9235	1.0	0.9567	1.0
12.0999	30.0	420	12.5604	1.0	0.9516	1.0
12.0999	31.0	434	10.6877	1.0	0.9469	1.0
12.0999	32.0	448	9.3758	1.0	0.9565	1.0
12.0999	33.0	462	5.0729	1.0	0.9619	1.0
12.0999	34.0	476	4.0927	1.0	0.9619	1.0
12.0999	35.0	490	3.8576	1.0	0.9619	1.0
7.922	36.0	504	3.7497	1.0	0.9619	1.0
7.922	37.0	518	3.6650	1.0	0.9619	1.0
7.922	38.0	532	3.5863	1.0	0.9619	1.0
7.922	39.0	546	3.5280	1.0	0.9619	1.0
7.922	40.0	560	3.4813	1.0	0.9619	1.0
7.922	41.0	574	3.4481	1.0	0.9619	1.0
7.922	42.0	588	3.4184	1.0	0.9619	1.0
3.5155	43.0	602	3.3964	1.0	0.9619	1.0
3.5155	44.0	616	3.3748	1.0	0.9619	1.0
3.5155	45.0	630	3.3545	1.0	0.9619	1.0
3.5155	46.0	644	3.3354	1.0	0.9619	1.0
3.5155	47.0	658	3.3090	1.0	0.9619	1.0
3.5155	48.0	672	3.2789	1.0	0.9619	1.0
3.5155	49.0	686	3.2441	1.0	0.9619	1.0
3.2278	50.0	700	3.2153	1.0	0.9619	1.0
3.2278	51.0	714	3.1921	1.0	0.9619	1.0
3.2278	52.0	728	3.1863	1.0	0.9619	1.0
3.2278	53.0	742	3.1605	1.0	0.9619	1.0
3.2278	54.0	756	3.1517	1.0	0.9619	1.0
3.2278	55.0	770	3.1389	1.0	0.9619	1.0
3.2278	56.0	784	3.1274	1.0	0.9619	1.0
3.2278	57.0	798	3.1237	1.0	0.9619	1.0
3.0881	58.0	812	3.1115	1.0	0.9619	1.0
3.0881	59.0	826	3.1051	1.0	0.9619	1.0
3.0881	60.0	840	3.1055	1.0	0.9619	1.0
3.0881	61.0	854	3.0982	1.0	0.9619	1.0
3.0881	62.0	868	3.0933	1.0	0.9619	1.0
3.0881	63.0	882	3.0871	1.0	0.9619	1.0
3.0881	64.0	896	3.0788	1.0	0.9619	1.0
3.0331	65.0	910	3.0835	1.0	0.9619	1.0
3.0331	66.0	924	3.0786	1.0	0.9619	1.0
3.0331	67.0	938	3.0781	1.0	0.9619	1.0
3.0331	68.0	952	3.0761	1.0	0.9619	1.0
3.0331	69.0	966	3.0663	1.0	0.9619	1.0
3.0331	70.0	980	3.0629	1.0	0.9619	1.0
3.0331	71.0	994	3.0661	1.0	0.9619	1.0
2.9941	72.0	1008	3.0600	1.0	0.9619	1.0
2.9941	73.0	1022	3.0559	1.0	0.9619	1.0
2.9941	74.0	1036	3.0517	1.0	0.9619	1.0
2.9941	75.0	1050	3.0524	1.0	0.9619	1.0
2.9941	76.0	1064	3.0506	1.0	0.9619	1.0
2.9941	77.0	1078	3.0451	1.0	0.9619	1.0
2.9941	78.0	1092	3.0485	1.0	0.9619	1.0
2.9748	79.0	1106	3.0472	1.0	0.9619	1.0
2.9748	80.0	1120	3.0464	1.0	0.9619	1.0
2.9748	81.0	1134	3.0458	1.0	0.9619	1.0
2.9748	82.0	1148	3.0386	1.0	0.9619	1.0
2.9748	83.0	1162	3.0376	1.0	0.9619	1.0
2.9748	84.0	1176	3.0365	1.0	0.9619	1.0
2.9748	85.0	1190	3.0414	1.0	0.9619	1.0
2.9573	86.0	1204	3.0400	1.0	0.9619	1.0
2.9573	87.0	1218	3.0327	1.0	0.9619	1.0
2.9573	88.0	1232	3.0354	1.0	0.9619	1.0
2.9573	89.0	1246	3.0313	1.0	0.9619	1.0
2.9573	90.0	1260	3.0344	1.0	0.9619	1.0
2.9573	91.0	1274	3.0385	1.0	0.9619	1.0
2.9573	92.0	1288	3.0343	1.0	0.9619	1.0
2.957	93.0	1302	3.0365	1.0	0.9619	1.0
2.957	94.0	1316	3.0292	1.0	0.9619	1.0
2.957	95.0	1330	3.0238	1.0	0.9619	1.0
2.957	96.0	1344	3.0332	1.0	0.9619	1.0
2.957	97.0	1358	3.0295	1.0	0.9619	1.0
2.957	98.0	1372	3.0305	1.0	0.9619	1.0
2.957	99.0	1386	3.0284	1.0	0.9619	1.0
2.9439	100.0	1400	3.0302	1.0	0.9619	1.0
2.9439	101.0	1414	3.0284	1.0	0.9619	1.0
2.9439	102.0	1428	3.0302	1.0	0.9619	1.0
2.9439	103.0	1442	3.0312	1.0	0.9619	1.0
2.9439	104.0	1456	3.0255	1.0	0.9619	1.0
2.9439	105.0	1470	3.0309	1.0	0.9619	1.0
2.9439	106.0	1484	3.0268	1.0	0.9619	1.0
2.9439	107.0	1498	3.0318	1.0	0.9619	1.0
2.9382	108.0	1512	3.0244	1.0	0.9619	1.0
2.9382	109.0	1526	3.0307	1.0	0.9619	1.0
2.9382	110.0	1540	3.0229	1.0	0.9619	1.0
2.9382	111.0	1554	3.0231	1.0	0.9619	1.0
2.9382	112.0	1568	3.0288	1.0	0.9619	1.0
2.9382	113.0	1582	3.0191	1.0	0.9619	1.0
2.9382	114.0	1596	3.0276	1.0	0.9619	1.0
2.9379	115.0	1610	3.0226	1.0	0.9619	1.0
2.9379	116.0	1624	3.0271	1.0	0.9619	1.0
2.9379	117.0	1638	3.0220	1.0	0.9619	1.0
2.9379	118.0	1652	3.0240	1.0	0.9619	1.0
2.9379	119.0	1666	3.0305	1.0	0.9619	1.0
2.9379	120.0	1680	3.0160	1.0	0.9619	1.0
2.9379	121.0	1694	3.0231	1.0	0.9619	1.0
2.9353	122.0	1708	3.0200	1.0	0.9619	1.0
2.9353	123.0	1722	3.0191	1.0	0.9619	1.0
2.9353	124.0	1736	3.0240	1.0	0.9619	1.0
2.9353	125.0	1750	3.0204	1.0	0.9619	1.0
2.9353	126.0	1764	3.0222	1.0	0.9619	1.0
2.9353	127.0	1778	3.0249	1.0	0.9619	1.0
2.9353	128.0	1792	3.0212	1.0	0.9619	1.0
2.9377	129.0	1806	3.0228	1.0	0.9619	1.0
2.9377	130.0	1820	3.0219	1.0	0.9619	1.0
2.9377	131.0	1834	3.0206	1.0	0.9619	1.0
2.9377	132.0	1848	3.0238	1.0	0.9619	1.0
2.9377	133.0	1862	3.0212	1.0	0.9619	1.0
2.9377	134.0	1876	3.0241	1.0	0.9619	1.0
2.9377	135.0	1890	3.0248	1.0	0.9619	1.0
2.929	136.0	1904	3.0250	1.0	0.9619	1.0
2.929	137.0	1918	3.0218	1.0	0.9619	1.0
2.929	138.0	1932	3.0230	1.0	0.9619	1.0
2.929	139.0	1946	3.0240	1.0	0.9619	1.0
2.929	140.0	1960	3.0226	1.0	0.9619	1.0

Framework versions

• Transformers 4.28.0
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.13.3