bert-suicide-detection-hk-large

This model is a fine-tuned version of wcyat/bert-suicide-detection-hk on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0695
• Accuracy: 0.9832

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.1588	0.0769	20	0.0662	0.9832
0.257	0.1538	40	0.0271	0.9916
0.1515	0.2308	60	0.0256	0.9832
0.1104	0.3077	80	0.0976	0.9580
0.062	0.3846	100	0.0559	0.9916
0.2219	0.4615	120	0.0380	0.9916
0.292	0.5385	140	0.1136	0.9748
0.0766	0.6154	160	0.0473	0.9916
0.1286	0.6923	180	0.0592	0.9916
0.0965	0.7692	200	0.0525	0.9832
0.1859	0.8462	220	0.0732	0.9748
0.0795	0.9231	240	0.0039	1.0
0.1486	1.0	260	0.0572	0.9832
0.0656	1.0769	280	0.0392	0.9916
0.001	1.1538	300	0.0501	0.9916
0.0014	1.2308	320	0.0973	0.9832
0.1297	1.3077	340	0.0905	0.9832
0.0004	1.3846	360	0.0639	0.9916
0.0387	1.4615	380	0.0674	0.9916
0.0551	1.5385	400	0.0661	0.9916
0.1413	1.6154	420	0.0660	0.9916
0.0004	1.6923	440	0.0663	0.9916
0.0514	1.7692	460	0.1180	0.9748
0.105	1.8462	480	0.0699	0.9832
0.0599	1.9231	500	0.1168	0.9748
0.0005	2.0	520	0.1709	0.9580
0.0007	2.0769	540	0.0688	0.9916
0.0002	2.1538	560	0.0670	0.9916
0.0002	2.2308	580	0.0673	0.9916
0.0001	2.3077	600	0.0683	0.9916
0.0001	2.3846	620	0.0688	0.9916
0.0002	2.4615	640	0.0699	0.9916
0.034	2.5385	660	0.1193	0.9748
0.1757	2.6154	680	0.0948	0.9748
0.0615	2.6923	700	0.0344	0.9916
0.0015	2.7692	720	0.0559	0.9916
0.0002	2.8462	740	0.0615	0.9916
0.0001	2.9231	760	0.0628	0.9916
0.0001	3.0	780	0.0635	0.9916
0.0001	3.0769	800	0.0643	0.9916
0.0001	3.1538	820	0.0648	0.9916
0.0001	3.2308	840	0.0654	0.9916
0.0001	3.3077	860	0.0661	0.9916
0.0005	3.3846	880	0.0670	0.9916
0.0006	3.4615	900	0.0682	0.9916
0.0695	3.5385	920	0.0669	0.9916
0.0001	3.6154	940	0.0656	0.9916
0.0372	3.6923	960	0.0632	0.9916
0.0802	3.7692	980	0.0546	0.9916
0.0002	3.8462	1000	0.0541	0.9916
0.0002	3.9231	1020	0.0561	0.9916
0.0098	4.0	1040	0.0601	0.9916
0.0002	4.0769	1060	0.0640	0.9916
0.0017	4.1538	1080	0.0682	0.9832
0.0001	4.2308	1100	0.0688	0.9916
0.0159	4.3077	1120	0.0669	0.9916
0.0001	4.3846	1140	0.0657	0.9916
0.0102	4.4615	1160	0.0676	0.9916
0.0327	4.5385	1180	0.0730	0.9832
0.0182	4.6154	1200	0.0717	0.9832
0.0001	4.6923	1220	0.0699	0.9832
0.0001	4.7692	1240	0.0698	0.9832
0.0001	4.8462	1260	0.0698	0.9832
0.0557	4.9231	1280	0.0695	0.9832
0.0347	5.0	1300	0.0695	0.9832

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1