layoutlmv3-wildreceipt

This model is a fine-tuned version of microsoft/layoutlmv3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3399
• Precision: 0.8693
• Recall: 0.8761
• F1: 0.8727
• Accuracy: 0.9225

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	0.4	100	1.3257	0.6259	0.2897	0.3961	0.6677
No log	0.8	200	0.8512	0.6479	0.5297	0.5829	0.7870
No log	1.2	300	0.6603	0.7311	0.6472	0.6866	0.8302
No log	1.6	400	0.5604	0.7580	0.7144	0.7355	0.8515
1.0344	2.0	500	0.4834	0.7920	0.7515	0.7712	0.8711
1.0344	2.4	600	0.4352	0.8070	0.7792	0.7929	0.8812
1.0344	2.8	700	0.4045	0.8146	0.8149	0.8148	0.8925
1.0344	3.2	800	0.4051	0.8067	0.8208	0.8137	0.8897
1.0344	3.6	900	0.3774	0.8211	0.8328	0.8269	0.8977
0.3942	4.0	1000	0.3556	0.8355	0.8340	0.8347	0.9025
0.3942	4.4	1100	0.3703	0.8211	0.8496	0.8351	0.9001
0.3942	4.8	1200	0.3430	0.8367	0.8486	0.8426	0.9057
0.3942	5.2	1300	0.3492	0.8349	0.8469	0.8409	0.9051
0.3942	5.6	1400	0.3259	0.8551	0.8498	0.8525	0.9108
0.2561	6.0	1500	0.3276	0.8422	0.8610	0.8515	0.9113
0.2561	6.4	1600	0.3307	0.8546	0.8497	0.8522	0.9110
0.2561	6.8	1700	0.3180	0.8527	0.8559	0.8543	0.9136
0.2561	7.2	1800	0.3239	0.8525	0.8593	0.8559	0.9135
0.2561	7.6	1900	0.3322	0.8499	0.8703	0.8600	0.9145
0.1866	8.0	2000	0.3265	0.8465	0.8681	0.8572	0.9131
0.1866	8.4	2100	0.3248	0.8588	0.8618	0.8603	0.9170
0.1866	8.8	2200	0.3269	0.8579	0.8629	0.8604	0.9162
0.1866	9.2	2300	0.3273	0.8656	0.8663	0.8660	0.9195
0.1866	9.6	2400	0.3312	0.8593	0.8702	0.8647	0.9187
0.1439	10.0	2500	0.3200	0.8639	0.8703	0.8671	0.9209
0.1439	10.4	2600	0.3367	0.8540	0.8761	0.8649	0.9183
0.1439	10.8	2700	0.3370	0.8614	0.8699	0.8656	0.9191
0.1439	11.2	2800	0.3294	0.8735	0.8690	0.8712	0.9221
0.1439	11.6	2900	0.3405	0.8653	0.8734	0.8693	0.9205
0.1186	12.0	3000	0.3334	0.8629	0.8767	0.8697	0.9210
0.1186	12.4	3100	0.3376	0.8653	0.8747	0.8700	0.9218
0.1186	12.8	3200	0.3362	0.8663	0.8752	0.8707	0.9209
0.1186	13.2	3300	0.3317	0.8729	0.8719	0.8724	0.9225
0.1186	13.6	3400	0.3408	0.8706	0.8714	0.8710	0.9213
0.0997	14.0	3500	0.3399	0.8693	0.8761	0.8727	0.9225
0.0997	14.4	3600	0.3445	0.8623	0.8767	0.8694	0.9208
0.0997	14.8	3700	0.3403	0.8673	0.8775	0.8724	0.9225
0.0997	15.2	3800	0.3492	0.8652	0.8763	0.8707	0.9209
0.0997	15.6	3900	0.3442	0.8692	0.8760	0.8726	0.9228
0.0891	16.0	4000	0.3441	0.8672	0.8767	0.8719	0.9225

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2