t5-small_stereoset_finetuned_HBRPOI

This model is a fine-tuned version of t5-small on the stereoset dataset. It achieves the following results on the evaluation set:

• Loss: 0.4383
• Accuracy: 0.6028
• Tp: 0.4890
• Tn: 0.1138
• Fp: 0.3854
• Fn: 0.0118

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: 0.0005
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Tp	Tn	Fp	Fn
0.4447	0.43	20	0.3978	0.5008	0.5008	0.0	0.4992	0.0
0.3776	0.85	40	0.3448	0.6232	0.5008	0.1224	0.3768	0.0
0.3649	1.28	60	0.3269	0.5612	0.5	0.0612	0.4380	0.0008
0.3275	1.7	80	0.3218	0.5330	0.4992	0.0338	0.4655	0.0016
0.2969	2.13	100	0.3104	0.6256	0.4961	0.1295	0.3697	0.0047
0.3283	2.55	120	0.3111	0.5730	0.4992	0.0738	0.4254	0.0016
0.3046	2.98	140	0.3040	0.5416	0.4992	0.0424	0.4568	0.0016
0.2603	3.4	160	0.3057	0.5447	0.4992	0.0455	0.4537	0.0016
0.2828	3.83	180	0.3186	0.5479	0.4984	0.0495	0.4498	0.0024
0.2326	4.26	200	0.3036	0.6193	0.4937	0.1256	0.3736	0.0071
0.2289	4.68	220	0.3328	0.5479	0.4976	0.0502	0.4490	0.0031
0.2234	5.11	240	0.3140	0.5777	0.4976	0.0801	0.4192	0.0031
0.2225	5.53	260	0.3245	0.5691	0.4976	0.0714	0.4278	0.0031
0.187	5.96	280	0.3300	0.5785	0.4961	0.0824	0.4168	0.0047
0.179	6.38	300	0.3344	0.5848	0.4961	0.0887	0.4105	0.0047
0.1523	6.81	320	0.3528	0.5895	0.4969	0.0926	0.4066	0.0039
0.1499	7.23	340	0.3788	0.6232	0.4906	0.1327	0.3666	0.0102
0.1292	7.66	360	0.3889	0.5942	0.4914	0.1028	0.3964	0.0094
0.13	8.09	380	0.3959	0.5903	0.4937	0.0965	0.4027	0.0071
0.1216	8.51	400	0.4169	0.5856	0.4922	0.0934	0.4058	0.0086
0.1306	8.94	420	0.4227	0.6005	0.4898	0.1107	0.3885	0.0110
0.0968	9.36	440	0.4334	0.5965	0.4914	0.1052	0.3940	0.0094
0.1044	9.79	460	0.4383	0.6028	0.4890	0.1138	0.3854	0.0118

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1
• Datasets 2.10.1
• Tokenizers 0.13.2