extra

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: nan
• Wer: 1.0
• Cer: 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: 0.0001
• 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_steps: 500
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
12.812	0.5025	100	5.3212	1.0	1.0
3.9783	1.0050	200	3.5194	1.0	1.0
3.1422	1.5075	300	3.0670	1.0	1.0
2.9862	2.0101	400	2.9835	1.0	1.0
2.9269	2.5126	500	2.9749	0.9998	0.9473
2.9034	3.0151	600	2.9679	1.0	0.9322
2.8646	3.5176	700	2.9003	1.0	0.9124
2.8467	4.0201	800	3.0267	0.9998	0.9322
2.8963	4.5226	900	2.9503	1.0	0.9812
2.8702	5.0251	1000	2.9332	1.0	0.9790
2.8738	5.5276	1100	2.9384	1.0	0.9780
2.8877	6.0302	1200	2.9477	1.0	0.9781
2.8648	6.5327	1300	2.9390	1.0	0.9762
2.8793	7.0352	1400	2.9702	1.0	0.9742
2.9068	7.5377	1500	2.9744	1.0	0.9740
2.9056	8.0402	1600	2.9744	1.0	0.9740
2.9093	8.5427	1700	2.9744	1.0	0.9740
2.9034	9.0452	1800	2.9744	1.0	0.9740
2.9031	9.5477	1900	2.9744	1.0	0.9740
2.9231	10.0503	2000	2.9744	1.0	0.9740
2.9061	10.5528	2100	2.9744	1.0	0.9740
2.9066	11.0553	2200	2.9744	1.0	0.9740
2.9031	11.5578	2300	2.9744	1.0	0.9740
2.9067	12.0603	2400	2.9744	1.0	0.9740
2.9062	12.5628	2500	2.9744	1.0	0.9740
2.9068	13.0653	2600	2.9744	1.0	0.9740
2.8895	13.5678	2700	2.9744	1.0	0.9740
2.9079	14.0704	2800	2.9744	1.0	0.9740
2.9034	14.5729	2900	2.9744	1.0	0.9740
2.9076	15.0754	3000	2.9744	1.0	0.9740
2.889	15.5779	3100	2.9744	1.0	0.9740
2.9085	16.0804	3200	2.9744	1.0	0.9740
2.9033	16.5829	3300	2.9744	1.0	0.9740
2.8952	17.0854	3400	2.9744	1.0	0.9740
2.903	17.5879	3500	2.9744	1.0	0.9740
2.9101	18.0905	3600	2.9744	1.0	0.9740
2.9167	18.5930	3700	2.9744	1.0	0.9740
2.9068	19.0955	3800	2.9744	1.0	0.9740
2.9062	19.5980	3900	2.9744	1.0	0.9740
2.9061	20.1005	4000	2.9744	1.0	0.9740
2.8916	20.6030	4100	2.9744	1.0	0.9740
2.9223	21.1055	4200	2.9744	1.0	0.9740
2.9046	21.6080	4300	2.9744	1.0	0.9740
2.9053	22.1106	4400	2.9744	1.0	0.9740
2.9052	22.6131	4500	2.9744	1.0	0.9740
2.9058	23.1156	4600	2.9744	1.0	0.9740
2.9068	23.6181	4700	2.9744	1.0	0.9740
2.9049	24.1206	4800	2.9744	1.0	0.9740
2.89	24.6231	4900	2.9744	1.0	0.9740
2.9065	25.1256	5000	2.9744	1.0	0.9740
2.9062	25.6281	5100	2.9744	1.0	0.9740
2.9052	26.1307	5200	2.9744	1.0	0.9740
2.8902	26.6332	5300	2.9744	1.0	0.9740
2.9091	27.1357	5400	2.9744	1.0	0.9740
2.9066	27.6382	5500	2.9744	1.0	0.9740
2.8929	28.1407	5600	2.9744	1.0	0.9740
2.9048	28.6432	5700	2.9744	1.0	0.9740
2.9027	29.1457	5800	2.9744	1.0	0.9740
2.903	29.6482	5900	2.9744	1.0	0.9740
2.908	30.1508	6000	2.9744	1.0	0.9740
2.9062	30.6533	6100	2.9744	1.0	0.9740
2.9062	31.1558	6200	2.9744	1.0	0.9740
2.893	31.6583	6300	2.9744	1.0	0.9740
2.9041	32.1608	6400	2.9744	1.0	0.9740
2.9051	32.6633	6500	2.9744	1.0	0.9740
2.9083	33.1658	6600	2.9744	1.0	0.9740
2.9043	33.6683	6700	2.9744	1.0	0.9740
2.905	34.1709	6800	2.9744	1.0	0.9740
2.9053	34.6734	6900	2.9744	1.0	0.9740
2.9047	35.1759	7000	2.9744	1.0	0.9740
2.8915	35.6784	7100	2.9744	1.0	0.9740
2.8923	36.1809	7200	2.9744	1.0	0.9740
2.8901	36.6834	7300	2.9744	1.0	0.9740
2.9089	37.1859	7400	2.9744	1.0	0.9740
4.6568	37.6884	7500	nan	1.0	1.0
0.0	38.1910	7600	nan	1.0	1.0
0.0	38.6935	7700	nan	1.0	1.0
0.0	39.1960	7800	nan	1.0	1.0
0.0	39.6985	7900	nan	1.0	1.0
0.0	40.2010	8000	nan	1.0	1.0
0.0	40.7035	8100	nan	1.0	1.0
0.0	41.2060	8200	nan	1.0	1.0
0.0	41.7085	8300	nan	1.0	1.0
0.0	42.2111	8400	nan	1.0	1.0
0.0	42.7136	8500	nan	1.0	1.0
0.0	43.2161	8600	nan	1.0	1.0
0.0	43.7186	8700	nan	1.0	1.0
0.0	44.2211	8800	nan	1.0	1.0
0.0	44.7236	8900	nan	1.0	1.0
0.0	45.2261	9000	nan	1.0	1.0
0.0	45.7286	9100	nan	1.0	1.0
0.0	46.2312	9200	nan	1.0	1.0
0.0	46.7337	9300	nan	1.0	1.0
0.0	47.2362	9400	nan	1.0	1.0
0.0	47.7387	9500	nan	1.0	1.0
0.0	48.2412	9600	nan	1.0	1.0
0.0	48.7437	9700	nan	1.0	1.0
0.0	49.2462	9800	nan	1.0	1.0
0.0	49.7487	9900	nan	1.0	1.0

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1