wh-ft-lr1e5-dtstf5-adm-ga1ba16-st15k-v2-evalstp10-pat20-trainvalch

This model is a fine-tuned version of HouraMor/wh-ft-lr5e6-dtstf5-adm-ga1ba16-st15k-v2-evalstp500-pat5 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5961
• Wer: 0.3040
• Cer: 0.2331

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
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.3637	0.0201	10	0.5637	0.3200	0.2353
0.3023	0.0402	20	0.5625	0.2982	0.2254
0.1974	0.0602	30	0.5612	0.2856	0.2150
0.3756	0.0803	40	0.5600	0.2797	0.2110
0.2867	0.1004	50	0.5607	0.2786	0.2099
0.259	0.1205	60	0.5629	0.2784	0.2106
0.2281	0.1406	70	0.5666	0.2792	0.2112
0.2824	0.1606	80	0.5690	0.3035	0.2323
0.2795	0.1807	90	0.5734	0.3068	0.2242
0.3335	0.2008	100	0.5738	0.3062	0.2240
0.2585	0.2209	110	0.5759	0.2824	0.2143
0.2231	0.2410	120	0.5806	0.3282	0.2520
0.2724	0.2610	130	0.5841	0.3354	0.2651
0.2431	0.2811	140	0.5914	0.2906	0.2210
0.3521	0.3012	150	0.5851	0.2781	0.2091
0.3887	0.3213	160	0.5820	0.2829	0.2116
0.2823	0.3414	170	0.5853	0.2907	0.2199
0.3459	0.3614	180	0.5939	0.2736	0.2095
0.177	0.3815	190	0.5961	0.3040	0.2331

Framework versions

• Transformers 4.55.2
• Pytorch 2.7.0+cu118
• Datasets 2.21.0
• Tokenizers 0.21.4