bert-suicide-detection-hk-new

This model is a fine-tuned version of hon9kon9ize/bert-base-cantonese on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3903
• Accuracy: 0.9333

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: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.4227	0.0615	20	0.3869	0.8267
0.4575	0.1231	40	0.2748	0.8733
0.4332	0.1846	60	0.2883	0.84
0.2946	0.2462	80	0.2482	0.8867
0.2335	0.3077	100	0.2182	0.8933
0.2751	0.3692	120	0.2767	0.9
0.327	0.4308	140	0.6645	0.8067
0.2839	0.4923	160	0.2197	0.9333
0.2436	0.5538	180	0.2382	0.9267
0.2855	0.6154	200	0.4087	0.88
0.3372	0.6769	220	0.2596	0.94
0.1343	0.7385	240	0.7997	0.84
0.285	0.8	260	0.3252	0.9067
0.145	0.8615	280	0.8378	0.8333
0.2577	0.9231	300	0.4026	0.9067
0.4514	0.9846	320	0.4263	0.8867
0.245	1.0462	340	0.3208	0.9067
0.0017	1.1077	360	0.5023	0.8733
0.0176	1.1692	380	0.5177	0.88
0.1223	1.2308	400	0.6029	0.88
0.1639	1.2923	420	0.6401	0.88
0.1752	1.3538	440	0.4151	0.9
0.1417	1.4154	460	0.2314	0.9467
0.1784	1.4769	480	0.4026	0.9133
0.1671	1.5385	500	0.4188	0.9067
0.2027	1.6	520	0.2420	0.94
0.1009	1.6615	540	0.5572	0.86
0.1411	1.7231	560	0.5484	0.8867
0.078	1.7846	580	0.2864	0.9333
0.2094	1.8462	600	0.4784	0.9067
0.2487	1.9077	620	0.2854	0.9267
0.1476	1.9692	640	0.2096	0.9467
0.0111	2.0308	660	0.3278	0.9333
0.056	2.0923	680	0.3028	0.94
0.0025	2.1538	700	0.4313	0.9
0.0171	2.2154	720	0.3401	0.9333
0.2359	2.2769	740	0.3079	0.9467
0.0966	2.3385	760	0.4836	0.9
0.0375	2.4	780	0.5409	0.88
0.1249	2.4615	800	0.2857	0.9467
0.0408	2.5231	820	0.2854	0.94
0.0685	2.5846	840	0.3301	0.94
0.0676	2.6462	860	0.4170	0.9067
0.09	2.7077	880	0.4455	0.9067
0.0011	2.7692	900	0.3954	0.9267
0.0198	2.8308	920	0.4213	0.9133
0.1061	2.8923	940	0.3032	0.94
0.0003	2.9538	960	0.3759	0.92
0.0003	3.0154	980	0.3952	0.92
0.0037	3.0769	1000	0.4295	0.9133
0.0003	3.1385	1020	0.4906	0.9133
0.0003	3.2	1040	0.4890	0.9133
0.0642	3.2615	1060	0.3462	0.9333
0.0003	3.3231	1080	0.3094	0.9467
0.0003	3.3846	1100	0.3282	0.94
0.1037	3.4462	1120	0.3809	0.9333
0.0006	3.5077	1140	0.4448	0.9267
0.0942	3.5692	1160	0.6031	0.8867
0.0003	3.6308	1180	0.4964	0.8867
0.0007	3.6923	1200	0.5269	0.8867
0.0887	3.7538	1220	0.4914	0.8867
0.0003	3.8154	1240	0.3959	0.9267
0.0008	3.8769	1260	0.4240	0.9267
0.0003	3.9385	1280	0.4334	0.92
0.0003	4.0	1300	0.4242	0.9267
0.0002	4.0615	1320	0.4218	0.9267
0.0003	4.1231	1340	0.4187	0.9267
0.0002	4.1846	1360	0.4103	0.9267
0.0002	4.2462	1380	0.4091	0.9267
0.0002	4.3077	1400	0.4111	0.9267
0.0003	4.3692	1420	0.4092	0.9267
0.0003	4.4308	1440	0.3991	0.9333
0.0002	4.4923	1460	0.3991	0.9333
0.0002	4.5538	1480	0.3986	0.9333
0.0004	4.6154	1500	0.4055	0.9333
0.1421	4.6769	1520	0.4006	0.9333
0.0002	4.7385	1540	0.4030	0.9267
0.0002	4.8	1560	0.4034	0.9267
0.0628	4.8615	1580	0.3876	0.9333
0.0003	4.9231	1600	0.3880	0.9333
0.0003	4.9846	1620	0.3903	0.9333

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0