donut-base-sroie-test

This model is a fine-tuned version of davelotito/donut-base-sroie-test on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.3913
• Bleu: 0.0706
• Precisions: [0.8125, 0.7440860215053764, 0.7064676616915423, 0.6637168141592921]
• Brevity Penalty: 0.0968
• Length Ratio: 0.2998
• Translation Length: 528
• Reference Length: 1761
• Cer: 0.7448
• Wer: 0.8259

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: 2e-05
• 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
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Precisions	Brevity Penalty	Length Ratio	Translation Length	Reference Length	Cer	Wer
No log	0.99	62	0.5515	0.0638	[0.7647058823529411, 0.6831896551724138, 0.6309226932668329, 0.5857988165680473]	0.0962	0.2993	527	1761	0.7627	0.8549
0.5624	2.0	125	0.4773	0.0665	[0.7763157894736842, 0.6865671641791045, 0.6403940886699507, 0.5918367346938775]	0.0992	0.3021	532	1761	0.7562	0.8390
0.5624	2.99	187	0.4273	0.0658	[0.7840909090909091, 0.6903225806451613, 0.6517412935323383, 0.6047197640117994]	0.0968	0.2998	528	1761	0.7513	0.8373
0.2964	4.0	250	0.4007	0.0679	[0.800376647834275, 0.7072649572649573, 0.6592592592592592, 0.6023391812865497]	0.0986	0.3015	531	1761	0.7478	0.8286
0.2238	4.99	312	0.3965	0.0710	[0.8142589118198874, 0.7297872340425532, 0.683046683046683, 0.6308139534883721]	0.0999	0.3027	533	1761	0.7427	0.8271
0.2238	6.0	375	0.3939	0.0719	[0.8301886792452831, 0.7537473233404711, 0.7054455445544554, 0.656891495601173]	0.0980	0.3010	530	1761	0.7414	0.8246
0.147	6.99	437	0.3853	0.0693	[0.8159392789373814, 0.7370689655172413, 0.6932668329177057, 0.6449704142011834]	0.0962	0.2993	527	1761	0.7437	0.8237
0.1316	8.0	500	0.3827	0.0698	[0.8037735849056604, 0.7301927194860813, 0.6856435643564357, 0.6392961876832844]	0.0980	0.3010	530	1761	0.7456	0.8296
0.1316	8.99	562	0.3895	0.0704	[0.8174904942965779, 0.7516198704103672, 0.715, 0.6706231454005934]	0.0956	0.2987	526	1761	0.7439	0.8259
0.1153	9.92	620	0.3913	0.0706	[0.8125, 0.7440860215053764, 0.7064676616915423, 0.6637168141592921]	0.0968	0.2998	528	1761	0.7448	0.8259

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.1.0
• Datasets 2.18.0
• Tokenizers 0.15.2