layoutlmv3-finetuned-wildreceipt

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

• Loss: 0.3068
• Precision: 0.8738
• Recall: 0.8809
• F1: 0.8774
• Accuracy: 0.9246

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	0.32	100	1.3498	0.6130	0.3126	0.4140	0.6742
No log	0.63	200	0.8939	0.6665	0.5317	0.5915	0.7815
No log	0.95	300	0.7159	0.7311	0.6425	0.6840	0.8161
No log	1.26	400	0.5901	0.7554	0.6690	0.7095	0.8405
1.0677	1.58	500	0.5263	0.7632	0.7232	0.7427	0.8578
1.0677	1.89	600	0.4759	0.7871	0.7777	0.7824	0.8774
1.0677	2.21	700	0.4299	0.8054	0.8070	0.8062	0.8890
1.0677	2.52	800	0.4165	0.8064	0.8311	0.8185	0.8937
1.0677	2.84	900	0.3845	0.8344	0.8300	0.8322	0.9005
0.4267	3.15	1000	0.3540	0.8433	0.8318	0.8375	0.9056
0.4267	3.47	1100	0.3429	0.8362	0.8540	0.8450	0.9086
0.4267	3.79	1200	0.3274	0.8451	0.8545	0.8498	0.9105
0.4267	4.1	1300	0.3433	0.8397	0.8535	0.8466	0.9092
0.4267	4.42	1400	0.3181	0.8514	0.8604	0.8559	0.9154
0.2869	4.73	1500	0.3191	0.8472	0.8637	0.8554	0.9129
0.2869	5.05	1600	0.3128	0.8613	0.8658	0.8635	0.9182
0.2869	5.36	1700	0.3121	0.8622	0.8695	0.8658	0.9182
0.2869	5.68	1800	0.3230	0.8473	0.8661	0.8566	0.9140
0.2869	5.99	1900	0.2986	0.8729	0.8633	0.8681	0.9209
0.2134	6.31	2000	0.3032	0.8555	0.8694	0.8624	0.9169
0.2134	6.62	2100	0.3056	0.8705	0.8710	0.8708	0.9220
0.2134	6.94	2200	0.3122	0.8630	0.8790	0.8709	0.9217
0.2134	7.26	2300	0.3047	0.8692	0.8778	0.8734	0.9215
0.2134	7.57	2400	0.3103	0.8701	0.8780	0.8741	0.9225
0.1661	7.89	2500	0.3080	0.8712	0.8787	0.8749	0.9226
0.1661	8.2	2600	0.3011	0.8653	0.8834	0.8743	0.9236
0.1661	8.52	2700	0.3034	0.8735	0.8798	0.8766	0.9247
0.1661	8.83	2800	0.3054	0.8698	0.8793	0.8745	0.9238
0.1661	9.15	2900	0.3105	0.8697	0.8812	0.8754	0.9237
0.1415	9.46	3000	0.3068	0.8738	0.8809	0.8774	0.9246
0.1415	9.78	3100	0.3086	0.8730	0.8793	0.8761	0.9229
0.1415	10.09	3200	0.3013	0.8755	0.8830	0.8792	0.9256
0.1415	10.41	3300	0.3107	0.8692	0.8815	0.8753	0.9241
0.1415	10.73	3400	0.3073	0.8759	0.8794	0.8777	0.9261
0.1239	11.04	3500	0.3109	0.8727	0.8819	0.8773	0.9253
0.1239	11.36	3600	0.3124	0.8723	0.8790	0.8756	0.9243
0.1239	11.67	3700	0.3171	0.8724	0.8805	0.8764	0.9241
0.1239	11.99	3800	0.3081	0.8739	0.8804	0.8771	0.9254
0.1239	12.3	3900	0.3095	0.8735	0.8798	0.8766	0.9254
0.1106	12.62	4000	0.3094	0.8740	0.8796	0.8768	0.9254

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 2.0.0
• Datasets 2.14.3
• Tokenizers 0.13.3