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

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

• Loss: 0.0131
• Semantic-accuracy: 0.36
• Gen Len: 18.24

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: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Semantic-accuracy	Gen Len
0.2382	1.0	563	0.0431	0.322	18.224
0.0477	2.0	1126	0.0229	0.356	18.236
0.0305	3.0	1689	0.0259	0.34	18.266
0.0231	4.0	2252	0.0204	0.35	18.238
0.0197	5.0	2815	0.0162	0.352	18.232
0.02	6.0	3378	0.0162	0.354	18.238
0.0172	7.0	3941	0.0147	0.356	18.24
0.0145	8.0	4504	0.0259	0.34	18.246
0.0133	9.0	5067	0.0129	0.358	18.238
0.0131	10.0	5630	0.0121	0.366	18.242
0.0122	11.0	6193	0.0128	0.354	18.242
0.0123	12.0	6756	0.0129	0.356	18.222
0.0113	13.0	7319	0.0131	0.362	18.232
0.0095	14.0	7882	0.0124	0.358	18.238
0.0102	15.0	8445	0.0127	0.362	18.244
0.0089	16.0	9008	0.0126	0.358	18.242
0.0086	17.0	9571	0.0133	0.358	18.242
0.0084	17.76	10000	0.0131	0.36	18.24

Framework versions

• Transformers 4.25.1
• Pytorch 1.12.1+cu113
• Datasets 2.7.1
• Tokenizers 0.13.2