ep-25-model

This model is a fine-tuned version of anvitamanne/base-model on the None dataset. It achieves the following results on the evaluation set:

• Loss: 577.1931
• Wer: 0.3901
• Cer: 0.1651

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 25
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
313.5326	0.86	1000	523.8419	0.4040	0.1657
312.9761	1.72	2000	510.8694	0.3953	0.1629
304.1435	2.58	3000	528.4339	0.3999	0.1662
298.9039	3.44	4000	511.5696	0.3985	0.1636
285.1419	4.3	5000	497.8684	0.3996	0.1639
284.229	5.17	6000	523.1895	0.3981	0.1645
277.664	6.03	7000	534.4621	0.3963	0.1647
279.9323	6.89	8000	533.2930	0.3988	0.1652
281.1303	7.75	9000	566.0540	0.4129	0.1721
265.8685	8.61	10000	532.5298	0.4045	0.1674
262.7262	9.47	11000	549.6611	0.4068	0.1687
253.1763	10.33	12000	569.0319	0.4080	0.1704
275.8374	11.19	13000	578.4818	0.4027	0.1703
259.9116	12.05	14000	575.9494	0.4039	0.1688
257.2954	12.91	15000	570.9092	0.4077	0.1699
262.4425	13.78	16000	563.3949	0.3944	0.1658
271.2236	14.64	17000	560.2855	0.3930	0.1652
271.719	15.5	18000	560.2625	0.3948	0.1660
273.512	16.36	19000	556.5314	0.3964	0.1667
269.4678	17.22	20000	561.3873	0.3949	0.1671
275.9994	18.08	21000	558.5574	0.3904	0.1646
269.7911	18.94	22000	553.4536	0.3982	0.1678
255.264	19.8	23000	569.4017	0.3919	0.1655
265.774	20.66	24000	571.5080	0.3906	0.1652
264.2804	21.52	25000	571.9384	0.3922	0.1657
268.9472	22.38	26000	575.9176	0.3890	0.1650
265.2463	23.25	27000	572.2623	0.3916	0.1657
266.2521	24.11	28000	575.5298	0.3895	0.1646
256.3034	24.97	29000	577.1931	0.3901	0.1651

Framework versions

• Transformers 4.38.2
• Pytorch 2.1.2+cu118
• Datasets 3.6.0
• Tokenizers 0.15.2