STS-conventional-Fine-Tuning-Capstone-roberta-base-filtered-170

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

• Loss: 2.3403
• Accuracy: 0.7285

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	113	0.7529	0.6816
No log	2.0	226	0.7985	0.7097
No log	3.0	339	0.8245	0.7097
No log	4.0	452	0.8816	0.6798
0.5011	5.0	565	1.0854	0.6929
0.5011	6.0	678	1.1921	0.7135
0.5011	7.0	791	1.3839	0.7228
0.5011	8.0	904	1.4560	0.7247
0.1649	9.0	1017	1.6387	0.7191
0.1649	10.0	1130	1.8012	0.7172
0.1649	11.0	1243	1.8790	0.7247
0.1649	12.0	1356	2.0223	0.7116
0.1649	13.0	1469	2.0297	0.7228
0.0639	14.0	1582	2.1202	0.7228
0.0639	15.0	1695	2.2489	0.7303
0.0639	16.0	1808	2.2505	0.7266
0.0639	17.0	1921	2.2693	0.7303
0.0198	18.0	2034	2.3216	0.7228
0.0198	19.0	2147	2.3244	0.7247
0.0198	20.0	2260	2.3403	0.7285

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2