paulc1212-nl-query-2-mongo-query-t5-small
This model is a fine-tuned version of t5-small on a 1138 examples dataset.
NOTE: It doesn't work on any mongodb and only on the trained one.
It achieves the following results on the evaluation set:
- Loss: 0.0039
Model description
The purpose of fine-tuning is to convert NL to mongodb queries.
Intended uses & limitations
It will work only for a particular mongodb (for which was fine-tune).
So it's not something generic that can handle any mongodb.
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.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: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.0341 | 1.0 | 254 | 0.0508 |
| 0.067 | 2.0 | 508 | 0.0295 |
| 0.045 | 3.0 | 762 | 0.0214 |
| 0.035 | 4.0 | 1016 | 0.0149 |
| 0.0275 | 5.0 | 1270 | 0.0116 |
| 0.0229 | 6.0 | 1524 | 0.0094 |
| 0.0201 | 7.0 | 1778 | 0.0081 |
| 0.0178 | 8.0 | 2032 | 0.0067 |
| 0.016 | 9.0 | 2286 | 0.0060 |
| 0.0143 | 10.0 | 2540 | 0.0054 |
| 0.0132 | 11.0 | 2794 | 0.0048 |
| 0.0127 | 12.0 | 3048 | 0.0045 |
| 0.0114 | 13.0 | 3302 | 0.0041 |
| 0.0113 | 14.0 | 3556 | 0.0040 |
| 0.0115 | 15.0 | 3810 | 0.0039 |
Framework versions
- Transformers 4.51.1
- Pytorch 2.6.0+cu124
- Datasets 2.14.4
- Tokenizers 0.21.1