lilt-en-funsd

KrisKrit/lilt-en-funsd

lilt-en-funsd is a token classification model from KrisKrit. Use it when you need labels on individual words, such as names. It is set up for transformers. The card lists the license as mit.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

4% of all-time downloads

All-time downloads

160

Public

Parameters

130M

9.4 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors521 MB · 99%

At a glance

Task: Token Classification
Library: transformers
License: mit
Model type: lilt
Access: Public
Created: Feb 21, 2024
Updated: Apr 3, 2024
SHA: c26dbcf1

Try a prompt

Base models

SCUT-DLVCLab/lilt-roberta-en-base

Model card

From the Hugging Face model README

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: 1.8012
Answer: {'precision': 0.8827751196172249, 'recall': 0.9033047735618115, 'f1': 0.8929219600725952, 'number': 817}
Header: {'precision': 0.6868686868686869, 'recall': 0.5714285714285714, 'f1': 0.6238532110091742, 'number': 119}
Question: {'precision': 0.9034296028880866, 'recall': 0.9294336118848654, 'f1': 0.9162471395881007, 'number': 1077}
Overall Precision: 0.8845
Overall Recall: 0.8977
Overall F1: 0.8910
Overall Accuracy: 0.8001

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
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
0.4051	10.53	200	1.1192	{'precision': 0.8408577878103838, 'recall': 0.9118727050183598, 'f1': 0.8749266001174398, 'number': 817}	{'precision': 0.44776119402985076, 'recall': 0.5042016806722689, 'f1': 0.4743083003952569, 'number': 119}	{'precision': 0.8896289248334919, 'recall': 0.8681522748375116, 'f1': 0.8787593984962405, 'number': 1077}	0.8402	0.8644	0.8521	0.7879
0.0479	21.05	400	1.3335	{'precision': 0.8625146886016452, 'recall': 0.8984088127294981, 'f1': 0.880095923261391, 'number': 817}	{'precision': 0.539568345323741, 'recall': 0.6302521008403361, 'f1': 0.5813953488372092, 'number': 119}	{'precision': 0.8969359331476323, 'recall': 0.8969359331476323, 'f1': 0.8969359331476322, 'number': 1077}	0.8587	0.8818	0.8701	0.7952
0.0135	31.58	600	1.5420	{'precision': 0.8301675977653631, 'recall': 0.9094247246022031, 'f1': 0.8679906542056074, 'number': 817}	{'precision': 0.6082474226804123, 'recall': 0.4957983193277311, 'f1': 0.5462962962962963, 'number': 119}	{'precision': 0.8853333333333333, 'recall': 0.924791086350975, 'f1': 0.9046321525885559, 'number': 1077}	0.8493	0.8932	0.8707	0.7890
0.0068	42.11	800	1.6315	{'precision': 0.8870967741935484, 'recall': 0.8751529987760098, 'f1': 0.8810844115834873, 'number': 817}	{'precision': 0.6521739130434783, 'recall': 0.6302521008403361, 'f1': 0.641025641025641, 'number': 119}	{'precision': 0.9054307116104869, 'recall': 0.8978644382544104, 'f1': 0.9016317016317017, 'number': 1077}	0.8834	0.8728	0.8781	0.8044
0.0047	52.63	1000	1.6176	{'precision': 0.8813559322033898, 'recall': 0.8910648714810282, 'f1': 0.8861838101034694, 'number': 817}	{'precision': 0.6764705882352942, 'recall': 0.5798319327731093, 'f1': 0.6244343891402716, 'number': 119}	{'precision': 0.8922528940338379, 'recall': 0.9303621169916435, 'f1': 0.9109090909090909, 'number': 1077}	0.8771	0.8937	0.8853	0.7979
0.002	63.16	1200	1.9150	{'precision': 0.8640552995391705, 'recall': 0.9179926560587516, 'f1': 0.8902077151335311, 'number': 817}	{'precision': 0.6330275229357798, 'recall': 0.5798319327731093, 'f1': 0.6052631578947367, 'number': 119}	{'precision': 0.9049815498154982, 'recall': 0.9108635097493036, 'f1': 0.907913003239241, 'number': 1077}	0.8734	0.8942	0.8837	0.7874
0.0015	73.68	1400	1.7303	{'precision': 0.8863361547762999, 'recall': 0.8971848225214198, 'f1': 0.8917274939172749, 'number': 817}	{'precision': 0.7319587628865979, 'recall': 0.5966386554621849, 'f1': 0.6574074074074073, 'number': 119}	{'precision': 0.8903985507246377, 'recall': 0.9127205199628597, 'f1': 0.901421366345713, 'number': 1077}	0.8812	0.8877	0.8844	0.8015
0.0011	84.21	1600	1.7743	{'precision': 0.8763376932223543, 'recall': 0.9020807833537332, 'f1': 0.8890229191797346, 'number': 817}	{'precision': 0.6147540983606558, 'recall': 0.6302521008403361, 'f1': 0.6224066390041495, 'number': 119}	{'precision': 0.9009259259259259, 'recall': 0.903435468895079, 'f1': 0.9021789522484933, 'number': 1077}	0.8737	0.8867	0.8802	0.7948
0.001	94.74	1800	1.8012	{'precision': 0.8827751196172249, 'recall': 0.9033047735618115, 'f1': 0.8929219600725952, 'number': 817}	{'precision': 0.6868686868686869, 'recall': 0.5714285714285714, 'f1': 0.6238532110091742, 'number': 119}	{'precision': 0.9034296028880866, 'recall': 0.9294336118848654, 'f1': 0.9162471395881007, 'number': 1077}	0.8845	0.8977	0.8910	0.8001
0.0005	105.26	2000	1.7293	{'precision': 0.8468571428571429, 'recall': 0.9069767441860465, 'f1': 0.875886524822695, 'number': 817}	{'precision': 0.696078431372549, 'recall': 0.5966386554621849, 'f1': 0.6425339366515838, 'number': 119}	{'precision': 0.9037580201649863, 'recall': 0.9155060352831941, 'f1': 0.9095940959409595, 'number': 1077}	0.8694	0.8932	0.8812	0.8066
0.0003	115.79	2200	1.7737	{'precision': 0.8584579976985041, 'recall': 0.9130966952264382, 'f1': 0.8849347568208777, 'number': 817}	{'precision': 0.6979166666666666, 'recall': 0.5630252100840336, 'f1': 0.6232558139534883, 'number': 119}	{'precision': 0.8930817610062893, 'recall': 0.9229340761374187, 'f1': 0.9077625570776255, 'number': 1077}	0.8696	0.8977	0.8834	0.8076
0.0002	126.32	2400	1.7984	{'precision': 0.8419864559819413, 'recall': 0.9130966952264382, 'f1': 0.8761009982384028, 'number': 817}	{'precision': 0.7040816326530612, 'recall': 0.5798319327731093, 'f1': 0.6359447004608296, 'number': 119}	{'precision': 0.9012797074954296, 'recall': 0.9155060352831941, 'f1': 0.9083371718102257, 'number': 1077}	0.8667	0.8947	0.8805	0.8078

Framework versions

Transformers 4.39.3
Pytorch 2.2.1+cu121
Datasets 2.18.0
Tokenizers 0.15.2

Task: Token Classification
Library: transformers
Type: lilt
License: mit
Created: Feb 21, 2024
Updated: Apr 3, 2024