NL-RX-Synth-t5-base-finetuned-en-to-regex

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

• Loss: 0.0255
• Semantic accuracy: 0.336
• Syntactic accuracy: 0.286
• Gen Len: 18.316

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: 0.001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Semantic accuracy	Syntactic accuracy	Gen Len
No log	0.18	100	0.2726	0.242	0.158	18.09
No log	0.36	200	0.1477	0.264	0.2	18.268
No log	0.53	300	0.1153	0.262	0.224	18.298
No log	0.71	400	0.0602	0.292	0.242	18.266
0.2992	0.89	500	0.0526	0.32	0.276	18.282
0.2992	1.07	600	0.0396	0.318	0.272	18.3
0.2992	1.24	700	0.0326	0.318	0.286	18.33
0.2992	1.42	800	0.0322	0.326	0.28	18.314
0.2992	1.6	900	0.0267	0.326	0.278	18.328
0.0383	1.78	1000	0.0255	0.336	0.286	18.316

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.0+cu116
• Datasets 2.7.1
• Tokenizers 0.13.2