TwHIN-BERT-Misinformation-Classifier

vickleso/TwHIN-BERT-Misinformation-Classifier

TwHIN-BERT-Misinformation-Classifier is a machine learning model from vickleso. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.

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

Downloads · 30 days

19% of all-time downloads

All-time downloads

Public

Parameters

561M

9 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.bin2.2 GB · 50%

Parameter types

How the weights are stored.

F32561M · 100%

Base models

Twitter/twhin-bert-large

Datasets

roupenminassian/twitter-misinformation

Model card

From the Hugging Face model README

TwHIN-BERT-Misinformation-Classifier

This model is a fine-tuned version of Twitter/twhin-bert-large on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.0632
F1: 0.9829

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: 2e-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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	F1
0.3092	0.04	250	0.0873	0.9683
0.095	0.09	500	0.0973	0.9752
0.0937	0.13	750	0.0969	0.9669
0.0859	0.17	1000	0.0762	0.9772
0.0828	0.22	1250	0.1208	0.9669
0.0685	0.26	1500	0.0745	0.9803
0.0801	0.3	1750	0.0678	0.9798
0.0717	0.35	2000	0.1027	0.9783
0.1132	0.39	2250	0.0968	0.9771
0.1088	0.43	2500	0.0838	0.9794
0.0976	0.48	2750	0.0915	0.9787
0.0863	0.52	3000	0.0828	0.9797
0.0967	0.56	3250	0.1350	0.9722
0.1225	0.61	3500	0.1098	0.9753
0.1017	0.65	3750	0.0807	0.9754
0.091	0.69	4000	0.1083	0.9757
0.0882	0.74	4250	0.0768	0.9811
0.0688	0.78	4500	0.0819	0.9788
0.0728	0.82	4750	0.0758	0.9802
0.0967	0.87	5000	0.1157	0.9743
0.0862	0.91	5250	0.0677	0.9816
0.0815	0.95	5500	0.0709	0.9768
0.0776	1.0	5750	0.0737	0.9799
0.0635	1.04	6000	0.0941	0.9761
0.0805	1.08	6250	0.0993	0.9778
0.0887	1.13	6500	0.0916	0.9770
0.0824	1.17	6750	0.0859	0.9790
0.0885	1.21	7000	0.0893	0.9795
0.0868	1.26	7250	0.0686	0.9801
0.09	1.3	7500	0.0940	0.9767
0.0907	1.34	7750	0.0878	0.9796
0.0835	1.39	8000	0.0736	0.9790
0.0747	1.43	8250	0.0878	0.9817
0.0745	1.47	8500	0.0675	0.9809
0.0754	1.52	8750	0.0708	0.9813
0.0781	1.56	9000	0.0816	0.9822
0.069	1.6	9250	0.0761	0.9820
0.0708	1.65	9500	0.0887	0.9814
0.1062	1.69	9750	0.0887	0.9803
0.0643	1.73	10000	0.0771	0.9821
0.0675	1.77	10250	0.0870	0.9809
0.0705	1.82	10500	0.0791	0.9817
0.0737	1.86	10750	0.0780	0.9815
0.0665	1.9	11000	0.0779	0.9828
0.0919	1.95	11250	0.0905	0.9768
0.0687	1.99	11500	0.0647	0.9806
0.0675	2.03	11750	0.0634	0.9814
0.0549	2.08	12000	0.0670	0.9812
0.0593	2.12	12250	0.0676	0.9815
0.0603	2.16	12500	0.0624	0.9828
0.0553	2.21	12750	0.0762	0.9806
0.0502	2.25	13000	0.0798	0.9835
0.0431	2.29	13250	0.1644	0.9227
0.0514	2.34	13500	0.0831	0.9803
0.0536	2.38	13750	0.0585	0.9833
0.0472	2.42	14000	0.0570	0.9839
0.0581	2.47	14250	0.0561	0.9828
0.0465	2.51	14500	0.0585	0.9842
0.0542	2.55	14750	0.1652	0.9827
0.0595	2.6	15000	0.1015	0.9821
0.051	2.64	15250	0.2099	0.8379
0.0447	2.68	15500	0.0633	0.9826
0.0389	2.73	15750	0.0647	0.9830
0.0533	2.77	16000	0.0623	0.9829
0.0524	2.81	16250	0.0586	0.9831
0.0576	2.86	16500	0.0597	0.9830
0.0483	2.9	16750	0.0689	0.9836
0.0469	2.94	17000	0.0569	0.9837
0.0447	2.99	17250	0.0632	0.9829

Framework versions

Transformers 4.28.1
Pytorch 2.0.0+cu118
Datasets 2.11.0
Tokenizers 0.13.3

Type: bert
License: apache-2.0
Created: Jun 23, 2026
Updated: Jun 23, 2026