layoutlm-funsd

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

• Loss: 0.6657
• Answer: {'precision': 0.7226519337016575, 'recall': 0.8084054388133498, 'f1': 0.763127187864644, 'number': 809}
• Header: {'precision': 0.29545454545454547, 'recall': 0.3277310924369748, 'f1': 0.3107569721115538, 'number': 119}
• Question: {'precision': 0.7931960608773501, 'recall': 0.831924882629108, 'f1': 0.8120989917506873, 'number': 1065}
• Overall Precision: 0.7331
• Overall Recall: 0.7923
• Overall F1: 0.7615
• Overall Accuracy: 0.8136

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.7892	1.0	10	1.6086	{'precision': 0.020948180815876516, 'recall': 0.023485784919653894, 'f1': 0.022144522144522148, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.20356472795497185, 'recall': 0.20375586854460093, 'f1': 0.20366025340215863, 'number': 1065}	0.1196	0.1184	0.1190	0.3742
1.4438	2.0	20	1.2175	{'precision': 0.22015915119363394, 'recall': 0.20519159456118666, 'f1': 0.2124120281509917, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.4544037412314887, 'recall': 0.5474178403755868, 'f1': 0.4965928449744464, 'number': 1065}	0.3677	0.3758	0.3717	0.5883
1.0512	3.0	30	0.9159	{'precision': 0.5192519251925193, 'recall': 0.5834363411619283, 'f1': 0.5494761350407451, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.6103327495621717, 'recall': 0.6544600938967137, 'f1': 0.6316266425011329, 'number': 1065}	0.5615	0.5866	0.5737	0.7102
0.8045	4.0	40	0.7549	{'precision': 0.6132264529058116, 'recall': 0.7564894932014833, 'f1': 0.6773657996679578, 'number': 809}	{'precision': 0.22, 'recall': 0.09243697478991597, 'f1': 0.13017751479289943, 'number': 119}	{'precision': 0.6795580110497238, 'recall': 0.6929577464788732, 'f1': 0.6861924686192468, 'number': 1065}	0.6378	0.6829	0.6596	0.7538
0.6559	5.0	50	0.6887	{'precision': 0.6546227417640808, 'recall': 0.761433868974042, 'f1': 0.704, 'number': 809}	{'precision': 0.25, 'recall': 0.16806722689075632, 'f1': 0.20100502512562815, 'number': 119}	{'precision': 0.6964285714285714, 'recall': 0.7323943661971831, 'f1': 0.7139588100686498, 'number': 1065}	0.6614	0.7105	0.6851	0.7764
0.547	6.0	60	0.6515	{'precision': 0.6659793814432989, 'recall': 0.7985166872682324, 'f1': 0.7262507026419337, 'number': 809}	{'precision': 0.2891566265060241, 'recall': 0.20168067226890757, 'f1': 0.23762376237623764, 'number': 119}	{'precision': 0.7140439932318104, 'recall': 0.7924882629107981, 'f1': 0.7512238540275923, 'number': 1065}	0.6774	0.7597	0.7162	0.7928
0.4923	7.0	70	0.6337	{'precision': 0.6784188034188035, 'recall': 0.7849196538936959, 'f1': 0.7277936962750717, 'number': 809}	{'precision': 0.2761904761904762, 'recall': 0.24369747899159663, 'f1': 0.2589285714285714, 'number': 119}	{'precision': 0.7371575342465754, 'recall': 0.8084507042253521, 'f1': 0.7711598746081505, 'number': 1065}	0.6904	0.7652	0.7258	0.8052
0.4463	8.0	80	0.6478	{'precision': 0.7045454545454546, 'recall': 0.7663782447466008, 'f1': 0.7341622261693309, 'number': 809}	{'precision': 0.2831858407079646, 'recall': 0.2689075630252101, 'f1': 0.27586206896551724, 'number': 119}	{'precision': 0.751937984496124, 'recall': 0.819718309859155, 'f1': 0.7843665768194069, 'number': 1065}	0.7080	0.7652	0.7355	0.8048
0.3974	9.0	90	0.6389	{'precision': 0.7029379760609358, 'recall': 0.7985166872682324, 'f1': 0.7476851851851851, 'number': 809}	{'precision': 0.2748091603053435, 'recall': 0.3025210084033613, 'f1': 0.288, 'number': 119}	{'precision': 0.7609254498714653, 'recall': 0.8338028169014085, 'f1': 0.7956989247311828, 'number': 1065}	0.7082	0.7878	0.7458	0.8060
0.3599	10.0	100	0.6429	{'precision': 0.7177777777777777, 'recall': 0.7985166872682324, 'f1': 0.7559976594499708, 'number': 809}	{'precision': 0.2966101694915254, 'recall': 0.29411764705882354, 'f1': 0.2953586497890296, 'number': 119}	{'precision': 0.7795275590551181, 'recall': 0.8366197183098592, 'f1': 0.8070652173913043, 'number': 1065}	0.7274	0.7888	0.7569	0.8139
0.3227	11.0	110	0.6510	{'precision': 0.710239651416122, 'recall': 0.8059332509270705, 'f1': 0.755066589461494, 'number': 809}	{'precision': 0.28205128205128205, 'recall': 0.2773109243697479, 'f1': 0.2796610169491525, 'number': 119}	{'precision': 0.7882037533512064, 'recall': 0.828169014084507, 'f1': 0.8076923076923077, 'number': 1065}	0.7275	0.7863	0.7557	0.8111
0.3156	12.0	120	0.6579	{'precision': 0.7245575221238938, 'recall': 0.8096415327564895, 'f1': 0.7647402218330415, 'number': 809}	{'precision': 0.2920353982300885, 'recall': 0.2773109243697479, 'f1': 0.28448275862068967, 'number': 119}	{'precision': 0.7926391382405745, 'recall': 0.8291079812206573, 'f1': 0.8104635153740247, 'number': 1065}	0.7372	0.7883	0.7619	0.8123
0.2935	13.0	130	0.6596	{'precision': 0.7119386637458927, 'recall': 0.8034610630407911, 'f1': 0.7549361207897795, 'number': 809}	{'precision': 0.2846715328467153, 'recall': 0.3277310924369748, 'f1': 0.3046875, 'number': 119}	{'precision': 0.7852112676056338, 'recall': 0.8375586854460094, 'f1': 0.8105406633348478, 'number': 1065}	0.7232	0.7933	0.7566	0.8131
0.2814	14.0	140	0.6629	{'precision': 0.7189901207464325, 'recall': 0.8096415327564895, 'f1': 0.7616279069767442, 'number': 809}	{'precision': 0.2923076923076923, 'recall': 0.31932773109243695, 'f1': 0.3052208835341365, 'number': 119}	{'precision': 0.7924528301886793, 'recall': 0.828169014084507, 'f1': 0.8099173553719008, 'number': 1065}	0.7312	0.7903	0.7596	0.8132
0.2762	15.0	150	0.6657	{'precision': 0.7226519337016575, 'recall': 0.8084054388133498, 'f1': 0.763127187864644, 'number': 809}	{'precision': 0.29545454545454547, 'recall': 0.3277310924369748, 'f1': 0.3107569721115538, 'number': 119}	{'precision': 0.7931960608773501, 'recall': 0.831924882629108, 'f1': 0.8120989917506873, 'number': 1065}	0.7331	0.7923	0.7615	0.8136

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3