wav2vec2-large-xlsr-coraa-exp-2

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: 0.7598
• Wer: 0.4437
• Cer: 0.2059

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: 3e-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
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
37.4939	1.0	14	27.8472	1.0	0.9612
37.4939	2.0	28	7.8236	1.0	0.9619
37.4939	3.0	42	4.5483	1.0	0.9619
37.4939	4.0	56	3.9455	1.0	0.9619
37.4939	5.0	70	3.6981	1.0	0.9619
37.4939	6.0	84	3.5551	1.0	0.9619
37.4939	7.0	98	3.4345	1.0	0.9619
9.4755	8.0	112	3.3129	1.0	0.9619
9.4755	9.0	126	3.2352	1.0	0.9619
9.4755	10.0	140	3.1795	1.0	0.9619
9.4755	11.0	154	3.1330	1.0	0.9619
9.4755	12.0	168	3.1329	1.0	0.9619
9.4755	13.0	182	3.0781	1.0	0.9619
9.4755	14.0	196	3.0682	1.0	0.9619
3.0804	15.0	210	3.0734	1.0	0.9619
3.0804	16.0	224	3.0565	1.0	0.9619
3.0804	17.0	238	3.0534	1.0	0.9619
3.0804	18.0	252	3.0327	1.0	0.9619
3.0804	19.0	266	3.0382	1.0	0.9619
3.0804	20.0	280	3.0320	1.0	0.9619
3.0804	21.0	294	3.0175	1.0	0.9619
2.9645	22.0	308	3.0184	1.0	0.9619
2.9645	23.0	322	3.0053	1.0	0.9619
2.9645	24.0	336	3.0082	1.0	0.9619
2.9645	25.0	350	3.0016	1.0	0.9619
2.9645	26.0	364	3.0093	1.0	0.9619
2.9645	27.0	378	2.9993	1.0	0.9619
2.9645	28.0	392	2.9998	1.0	0.9619
2.9326	29.0	406	3.0039	1.0	0.9619
2.9326	30.0	420	3.0013	1.0	0.9619
2.9326	31.0	434	2.9991	1.0	0.9619
2.9326	32.0	448	2.9949	1.0	0.9619
2.9326	33.0	462	2.9920	1.0	0.9619
2.9326	34.0	476	2.9980	1.0	0.9619
2.9326	35.0	490	2.9905	1.0	0.9619
2.9228	36.0	504	2.9935	1.0	0.9619
2.9228	37.0	518	2.9859	1.0	0.9619
2.9228	38.0	532	2.9879	1.0	0.9619
2.9228	39.0	546	2.9838	1.0	0.9619
2.9228	40.0	560	2.9819	1.0	0.9619
2.9228	41.0	574	2.9832	1.0	0.9619
2.9228	42.0	588	2.9748	1.0	0.9619
2.9107	43.0	602	2.9705	1.0	0.9616
2.9107	44.0	616	2.9658	1.0	0.9591
2.9107	45.0	630	2.9676	1.0	0.9580
2.9107	46.0	644	2.9602	1.0	0.9617
2.9107	47.0	658	2.9288	1.0	0.9581
2.9107	48.0	672	2.9023	1.0	0.9564
2.9107	49.0	686	2.8598	1.0	0.9601
2.8675	50.0	700	2.8022	1.0	0.9617
2.8675	51.0	714	2.7749	1.0	0.9559
2.8675	52.0	728	2.7368	1.0	0.9614
2.8675	53.0	742	2.6779	1.0	0.9597
2.8675	54.0	756	2.6466	1.0	0.9513
2.8675	55.0	770	2.6083	1.0	0.9381
2.8675	56.0	784	2.5400	1.0	0.9046
2.8675	57.0	798	2.4022	1.0	0.8121
2.6433	58.0	812	2.2463	1.0	0.7267
2.6433	59.0	826	2.0689	1.0	0.6162
2.6433	60.0	840	1.9145	1.0	0.5704
2.6433	61.0	854	1.7756	1.0	0.5095
2.6433	62.0	868	1.6238	1.0	0.4700
2.6433	63.0	882	1.4970	1.0	0.4454
2.6433	64.0	896	1.4010	1.0	0.4264
2.0023	65.0	910	1.3292	1.0	0.4142
2.0023	66.0	924	1.2790	0.9996	0.4043
2.0023	67.0	938	1.2129	0.9972	0.3898
2.0023	68.0	952	1.1590	0.9937	0.3795
2.0023	69.0	966	1.1193	0.9793	0.3618
2.0023	70.0	980	1.0872	0.9567	0.3482
2.0023	71.0	994	1.0603	0.9025	0.3278
1.3694	72.0	1008	1.0181	0.8694	0.3145
1.3694	73.0	1022	0.9941	0.8249	0.3000
1.3694	74.0	1036	0.9689	0.7082	0.2688
1.3694	75.0	1050	0.9346	0.6274	0.2466
1.3694	76.0	1064	0.9144	0.5603	0.2331
1.3694	77.0	1078	0.8997	0.5238	0.2253
1.3694	78.0	1092	0.8630	0.5154	0.2227
1.0996	79.0	1106	0.8602	0.4974	0.2171
1.0996	80.0	1120	0.8447	0.4911	0.2166
1.0996	81.0	1134	0.8308	0.4890	0.2164
1.0996	82.0	1148	0.8408	0.4781	0.2146
1.0996	83.0	1162	0.8201	0.4732	0.2128
1.0996	84.0	1176	0.8140	0.4711	0.2120
1.0996	85.0	1190	0.8041	0.4655	0.2105
0.9419	86.0	1204	0.7987	0.4616	0.2104
0.9419	87.0	1218	0.7883	0.4588	0.2092
0.9419	88.0	1232	0.7889	0.4581	0.2093
0.9419	89.0	1246	0.7859	0.4553	0.2088
0.9419	90.0	1260	0.7796	0.4508	0.2079
0.9419	91.0	1274	0.7814	0.4496	0.2081
0.9419	92.0	1288	0.7753	0.4482	0.2079
0.8707	93.0	1302	0.7674	0.4480	0.2068
0.8707	94.0	1316	0.7664	0.4474	0.2065
0.8707	95.0	1330	0.7612	0.4478	0.2063
0.8707	96.0	1344	0.7640	0.4441	0.2062
0.8707	97.0	1358	0.7598	0.4437	0.2059
0.8707	98.0	1372	0.7621	0.4427	0.2058
0.8707	99.0	1386	0.7634	0.4427	0.2059
0.8144	100.0	1400	0.7631	0.4425	0.2059

Framework versions

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