lilt-en-funsd

This model is a fine-tuned version of SCUT-DLVCLab/lilt-roberta-en-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6374
• Answer: {'precision': 0.86558516801854, 'recall': 0.9143206854345165, 'f1': 0.8892857142857143, 'number': 817}
• Header: {'precision': 0.6020408163265306, 'recall': 0.4957983193277311, 'f1': 0.543778801843318, 'number': 119}
• Question: {'precision': 0.8907788719785139, 'recall': 0.9238625812441968, 'f1': 0.9070191431175934, 'number': 1077}
• Overall Precision: 0.8667
• Overall Recall: 0.8947
• Overall F1: 0.8805
• Overall Accuracy: 0.8114

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• training_steps: 2500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.3924	10.5263	200	1.0090	{'precision': 0.8406779661016949, 'recall': 0.9106487148102815, 'f1': 0.8742655699177438, 'number': 817}	{'precision': 0.463768115942029, 'recall': 0.5378151260504201, 'f1': 0.49805447470817127, 'number': 119}	{'precision': 0.8781630740393627, 'recall': 0.8700092850510678, 'f1': 0.8740671641791045, 'number': 1077}	0.8349	0.8669	0.8506	0.7882
0.0455	21.0526	400	1.4106	{'precision': 0.8487972508591065, 'recall': 0.9069767441860465, 'f1': 0.8769230769230769, 'number': 817}	{'precision': 0.5384615384615384, 'recall': 0.5882352941176471, 'f1': 0.5622489959839357, 'number': 119}	{'precision': 0.8902104300091491, 'recall': 0.903435468895079, 'f1': 0.8967741935483872, 'number': 1077}	0.8511	0.8862	0.8683	0.7952
0.0133	31.5789	600	1.5157	{'precision': 0.8369942196531792, 'recall': 0.8861689106487148, 'f1': 0.8608799048751485, 'number': 817}	{'precision': 0.704225352112676, 'recall': 0.42016806722689076, 'f1': 0.5263157894736842, 'number': 119}	{'precision': 0.8806509945750453, 'recall': 0.904363974001857, 'f1': 0.8923499770957399, 'number': 1077}	0.8560	0.8684	0.8621	0.7963
0.0069	42.1053	800	1.6133	{'precision': 0.8629500580720093, 'recall': 0.9094247246022031, 'f1': 0.8855780691299165, 'number': 817}	{'precision': 0.5083333333333333, 'recall': 0.5126050420168067, 'f1': 0.5104602510460251, 'number': 119}	{'precision': 0.8908256880733945, 'recall': 0.9015784586815228, 'f1': 0.8961698200276881, 'number': 1077}	0.8571	0.8818	0.8692	0.7976
0.0032	52.6316	1000	1.8274	{'precision': 0.8084415584415584, 'recall': 0.9143206854345165, 'f1': 0.8581275129236072, 'number': 817}	{'precision': 0.6551724137931034, 'recall': 0.4789915966386555, 'f1': 0.5533980582524272, 'number': 119}	{'precision': 0.8783542039355993, 'recall': 0.9117920148560817, 'f1': 0.894760820045558, 'number': 1077}	0.8389	0.8872	0.8624	0.7852
0.0037	63.1579	1200	1.5619	{'precision': 0.8635294117647059, 'recall': 0.8984088127294981, 'f1': 0.8806238752249551, 'number': 817}	{'precision': 0.5357142857142857, 'recall': 0.6302521008403361, 'f1': 0.5791505791505792, 'number': 119}	{'precision': 0.8942486085343229, 'recall': 0.8950789229340761, 'f1': 0.8946635730858469, 'number': 1077}	0.8574	0.8808	0.8689	0.8029
0.0016	73.6842	1400	1.5773	{'precision': 0.8776978417266187, 'recall': 0.8959608323133414, 'f1': 0.8867353119321623, 'number': 817}	{'precision': 0.5887850467289719, 'recall': 0.5294117647058824, 'f1': 0.5575221238938053, 'number': 119}	{'precision': 0.8938700823421775, 'recall': 0.9071494893221913, 'f1': 0.9004608294930875, 'number': 1077}	0.8712	0.8803	0.8757	0.8164
0.0011	84.2105	1600	1.6210	{'precision': 0.8524404086265607, 'recall': 0.9192166462668299, 'f1': 0.8845700824499411, 'number': 817}	{'precision': 0.6588235294117647, 'recall': 0.47058823529411764, 'f1': 0.5490196078431372, 'number': 119}	{'precision': 0.8979033728350045, 'recall': 0.914577530176416, 'f1': 0.9061637534498619, 'number': 1077}	0.8686	0.8902	0.8793	0.8105
0.0005	94.7368	1800	1.6534	{'precision': 0.875886524822695, 'recall': 0.9069767441860465, 'f1': 0.8911605532170775, 'number': 817}	{'precision': 0.5739130434782609, 'recall': 0.5546218487394958, 'f1': 0.5641025641025642, 'number': 119}	{'precision': 0.8945454545454545, 'recall': 0.9136490250696379, 'f1': 0.9039963252181902, 'number': 1077}	0.8690	0.8897	0.8792	0.8078
0.0005	105.2632	2000	1.6261	{'precision': 0.8844765342960289, 'recall': 0.8996328029375765, 'f1': 0.8919902912621359, 'number': 817}	{'precision': 0.5803571428571429, 'recall': 0.5462184873949579, 'f1': 0.5627705627705628, 'number': 119}	{'precision': 0.879295154185022, 'recall': 0.9266480965645311, 'f1': 0.9023508137432187, 'number': 1077}	0.8653	0.8932	0.8790	0.8164
0.0006	115.7895	2200	1.6545	{'precision': 0.8589449541284404, 'recall': 0.9167686658506732, 'f1': 0.8869153345174661, 'number': 817}	{'precision': 0.5849056603773585, 'recall': 0.5210084033613446, 'f1': 0.5511111111111111, 'number': 119}	{'precision': 0.8871841155234657, 'recall': 0.9127205199628597, 'f1': 0.8997711670480548, 'number': 1077}	0.8600	0.8912	0.8753	0.8049
0.0003	126.3158	2400	1.6374	{'precision': 0.86558516801854, 'recall': 0.9143206854345165, 'f1': 0.8892857142857143, 'number': 817}	{'precision': 0.6020408163265306, 'recall': 0.4957983193277311, 'f1': 0.543778801843318, 'number': 119}	{'precision': 0.8907788719785139, 'recall': 0.9238625812441968, 'f1': 0.9070191431175934, 'number': 1077}	0.8667	0.8947	0.8805	0.8114

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.0