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: 0.7011
• Answer: {'precision': 0.7142857142857143, 'recall': 0.8096415327564895, 'f1': 0.7589803012746235, 'number': 809}
• Header: {'precision': 0.2962962962962963, 'recall': 0.33613445378151263, 'f1': 0.31496062992125984, 'number': 119}
• Question: {'precision': 0.7859712230215827, 'recall': 0.8206572769953052, 'f1': 0.8029398254478639, 'number': 1065}
• Overall Precision: 0.7250
• Overall Recall: 0.7873
• Overall F1: 0.7549
• Overall Accuracy: 0.8102

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

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.7566	1.0	10	1.5349	{'precision': 0.03646308113035551, 'recall': 0.049443757725587144, 'f1': 0.04197271773347323, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.16700819672131148, 'recall': 0.15305164319248826, 'f1': 0.15972562469377757, 'number': 1065}	0.0979	0.1019	0.0999	0.4336
1.4057	2.0	20	1.1865	{'precision': 0.17656500802568217, 'recall': 0.13597033374536466, 'f1': 0.15363128491620115, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.471847739888977, 'recall': 0.5586854460093896, 'f1': 0.5116079105760963, 'number': 1065}	0.3742	0.3537	0.3637	0.6016
1.0729	3.0	30	0.9241	{'precision': 0.49693251533742333, 'recall': 0.5006180469715699, 'f1': 0.4987684729064039, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.6378708551483421, 'recall': 0.6863849765258216, 'f1': 0.6612392582541836, 'number': 1065}	0.5691	0.5700	0.5696	0.7181
0.8134	4.0	40	0.7831	{'precision': 0.6211640211640211, 'recall': 0.7255871446229913, 'f1': 0.669327251995439, 'number': 809}	{'precision': 0.09375, 'recall': 0.05042016806722689, 'f1': 0.0655737704918033, 'number': 119}	{'precision': 0.6889081455805892, 'recall': 0.7464788732394366, 'f1': 0.7165389815232085, 'number': 1065}	0.6417	0.6964	0.6679	0.7640
0.6582	5.0	50	0.7298	{'precision': 0.6422018348623854, 'recall': 0.7787391841779975, 'f1': 0.7039106145251396, 'number': 809}	{'precision': 0.2361111111111111, 'recall': 0.14285714285714285, 'f1': 0.17801047120418848, 'number': 119}	{'precision': 0.7311233885819521, 'recall': 0.7455399061032864, 'f1': 0.7382612738261274, 'number': 1065}	0.6737	0.7230	0.6975	0.7761
0.553	6.0	60	0.6763	{'precision': 0.6673532440782698, 'recall': 0.8009888751545118, 'f1': 0.7280898876404494, 'number': 809}	{'precision': 0.25806451612903225, 'recall': 0.20168067226890757, 'f1': 0.22641509433962265, 'number': 119}	{'precision': 0.735445205479452, 'recall': 0.8065727699530516, 'f1': 0.7693685624720108, 'number': 1065}	0.6859	0.7682	0.7247	0.7962
0.4805	7.0	70	0.6797	{'precision': 0.6904255319148936, 'recall': 0.8022249690976514, 'f1': 0.7421383647798742, 'number': 809}	{'precision': 0.25925925925925924, 'recall': 0.23529411764705882, 'f1': 0.24669603524229072, 'number': 119}	{'precision': 0.7363945578231292, 'recall': 0.8131455399061033, 'f1': 0.7728692547969657, 'number': 1065}	0.6938	0.7742	0.7318	0.7970
0.4259	8.0	80	0.6726	{'precision': 0.689401888772298, 'recall': 0.8121137206427689, 'f1': 0.7457434733257663, 'number': 809}	{'precision': 0.24786324786324787, 'recall': 0.24369747899159663, 'f1': 0.24576271186440676, 'number': 119}	{'precision': 0.7463581833761782, 'recall': 0.8178403755868544, 'f1': 0.7804659498207885, 'number': 1065}	0.6960	0.7812	0.7362	0.8020
0.3787	9.0	90	0.6784	{'precision': 0.7043956043956044, 'recall': 0.792336217552534, 'f1': 0.7457824316463061, 'number': 809}	{'precision': 0.26229508196721313, 'recall': 0.2689075630252101, 'f1': 0.26556016597510373, 'number': 119}	{'precision': 0.779707495429616, 'recall': 0.8009389671361502, 'f1': 0.7901806391848076, 'number': 1065}	0.7178	0.7657	0.7410	0.8026
0.3411	10.0	100	0.6821	{'precision': 0.7015086206896551, 'recall': 0.8046971569839307, 'f1': 0.7495682210708117, 'number': 809}	{'precision': 0.2708333333333333, 'recall': 0.3277310924369748, 'f1': 0.2965779467680608, 'number': 119}	{'precision': 0.775200713648528, 'recall': 0.815962441314554, 'f1': 0.7950594693504116, 'number': 1065}	0.7109	0.7822	0.7449	0.8047
0.313	11.0	110	0.7129	{'precision': 0.7111111111111111, 'recall': 0.7911001236093943, 'f1': 0.7489760093622002, 'number': 809}	{'precision': 0.2835820895522388, 'recall': 0.31932773109243695, 'f1': 0.30039525691699603, 'number': 119}	{'precision': 0.7816711590296496, 'recall': 0.8169014084507042, 'f1': 0.7988980716253444, 'number': 1065}	0.7210	0.7767	0.7478	0.7994
0.297	12.0	120	0.6955	{'precision': 0.708779443254818, 'recall': 0.8182941903584673, 'f1': 0.759609868043603, 'number': 809}	{'precision': 0.291044776119403, 'recall': 0.3277310924369748, 'f1': 0.308300395256917, 'number': 119}	{'precision': 0.783978397839784, 'recall': 0.8178403755868544, 'f1': 0.8005514705882352, 'number': 1065}	0.7214	0.7888	0.7536	0.8103
0.2907	13.0	130	0.7098	{'precision': 0.7092511013215859, 'recall': 0.796044499381953, 'f1': 0.7501456027955737, 'number': 809}	{'precision': 0.3142857142857143, 'recall': 0.3697478991596639, 'f1': 0.33976833976833976, 'number': 119}	{'precision': 0.7896678966789668, 'recall': 0.8037558685446009, 'f1': 0.796649604467194, 'number': 1065}	0.7242	0.7747	0.7486	0.8052
0.2701	14.0	140	0.7006	{'precision': 0.7133479212253829, 'recall': 0.8059332509270705, 'f1': 0.7568195008705745, 'number': 809}	{'precision': 0.3037037037037037, 'recall': 0.3445378151260504, 'f1': 0.3228346456692913, 'number': 119}	{'precision': 0.7894736842105263, 'recall': 0.8169014084507042, 'f1': 0.8029533917858791, 'number': 1065}	0.7266	0.7842	0.7543	0.8091
0.2649	15.0	150	0.7011	{'precision': 0.7142857142857143, 'recall': 0.8096415327564895, 'f1': 0.7589803012746235, 'number': 809}	{'precision': 0.2962962962962963, 'recall': 0.33613445378151263, 'f1': 0.31496062992125984, 'number': 119}	{'precision': 0.7859712230215827, 'recall': 0.8206572769953052, 'f1': 0.8029398254478639, 'number': 1065}	0.7250	0.7873	0.7549	0.8102

Framework versions

• Transformers 4.32.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3