macarious/torgo_xlsr_finetune_F01_old

torgo_xlsr_finetune_F01

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.5082
• Wer: 0.2438

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.5402	0.6	1000	3.3380	1.0
2.2307	1.2	2000	1.8396	0.8959
1.0275	1.79	3000	1.5920	0.6564
0.7496	2.39	4000	1.6012	0.5640
0.6224	2.99	5000	1.3718	0.4902
0.5017	3.59	6000	1.8025	0.4902
0.4519	4.19	7000	1.4430	0.4017
0.4049	4.78	8000	1.6817	0.3748
0.3936	5.38	9000	1.5499	0.4069
0.384	5.98	10000	1.3839	0.3488
0.3331	6.58	11000	1.4815	0.3614
0.3168	7.18	12000	1.6359	0.3636
0.3262	7.78	13000	1.4287	0.3540
0.2877	8.37	14000	1.3445	0.3132
0.2684	8.97	15000	1.2778	0.2911
0.2306	9.57	16000	1.5658	0.3033
0.2215	10.17	17000	1.6831	0.2959
0.2381	10.77	18000	1.4930	0.2798
0.2237	11.36	19000	1.4210	0.2985
0.2227	11.96	20000	1.5231	0.2950
0.1965	12.56	21000	1.5101	0.2777
0.187	13.16	22000	1.6570	0.2998
0.1974	13.76	23000	1.4534	0.2772
0.1816	14.35	24000	1.6751	0.2755
0.1753	14.95	25000	1.5616	0.2707
0.1634	15.55	26000	1.4873	0.2469
0.1405	16.15	27000	1.6986	0.2590
0.1515	16.75	28000	1.5311	0.2477
0.1532	17.34	29000	1.3262	0.2403
0.1541	17.94	30000	1.3799	0.2477
0.1386	18.54	31000	1.4252	0.2408
0.1389	19.14	32000	1.5260	0.2434
0.1164	19.74	33000	1.5082	0.2438

Framework versions

• Transformers 4.26.1
• Pytorch 2.2.0
• Datasets 2.16.1
• Tokenizers 0.13.3