SetFit with sentence-transformers/all-mpnet-base-v2
This is a SetFit model trained on the konsman/setfit-messages-updated-influence-level dataset that can be used for Text Classification. This SetFit model uses sentence-transformers/all-mpnet-base-v2 as the Sentence Transformer embedding model. A LogisticRegression instance is used for classification.
The model has been trained using an efficient few-shot learning technique that involves:
- Fine-tuning a Sentence Transformer with contrastive learning.
- Training a classification head with features from the fine-tuned Sentence Transformer.
Model Details
Model Description
Model Sources
Model Labels
| Label |
Examples |
| 0 |
- 'The influence level of Understanding the effects of aging on the body is key for caregivers.'
- 'The influence level of Regular check-ups are key to maintaining good health.'
- 'The influence level of Balanced nutrition is key for maintaining health in old age.'
|
| 1 |
- "The influence level of Time for your 3pm medication! Please take as directed. Friendly reminder: It's time for your 3pm medication. Ensure to take it as prescribed."
- 'The influence level of Regular bladder function tests are important for elderly individuals.'
- 'The influence level of How was your telehealth session? Share your feedback. Help us improve. Provide feedback on your recent telehealth appointment. '
|
| 2 |
- "The influence level of A support group meeting is scheduled for tomorrow at 5pm. It's a great opportunity to share and learn. Connect with others in our support group meeting tomorrow. See you at 5pm!"
- 'The influence level of Safety first! Please update your emergency contact details in our system. Ensure swift help when needed. Update your emergency contacts in our app. '
- 'The influence level of Regularly discussing health concerns with doctors is important for the elderly.'
|
| 3 |
- 'The influence level of Understanding the role of dietary supplements in elderly health is important.'
- 'The influence level of Proper medication management is essential for effective treatment.'
- "The influence level of Your child's health is paramount. Reminder for the pediatrician appointment tomorrow. Ensure the best for your child. Don't miss the pediatrician appointment set for tomorrow. "
|
Evaluation
Metrics
| Label |
Accuracy |
| all |
0.4737 |
Uses
Direct Use for Inference
First install the SetFit library:
pip install setfit
Then you can load this model and run inference.
from setfit import SetFitModel
model = SetFitModel.from_pretrained("konsman/setfit-messages-label-v2")
preds = model("The influence level of Regularly updating emergency contact information is important for the elderly.")
Training Details
Training Set Metrics
| Training set |
Min |
Median |
Max |
| Word count |
12 |
20.8438 |
36 |
| Label |
Training Sample Count |
| 0 |
8 |
| 1 |
8 |
| 2 |
8 |
| 3 |
8 |
Training Hyperparameters
- batch_size: (8, 8)
- num_epochs: (4, 4)
- max_steps: -1
- sampling_strategy: oversampling
- num_iterations: 40
- body_learning_rate: (2.2041595048800003e-05, 2.2041595048800003e-05)
- head_learning_rate: 2.2041595048800003e-05
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: False
- use_amp: False
- warmup_proportion: 0.1
- seed: 42
- eval_max_steps: -1
- load_best_model_at_end: False
Training Results
| Epoch |
Step |
Training Loss |
Validation Loss |
| 0.0031 |
1 |
0.1587 |
- |
| 0.1562 |
50 |
0.116 |
- |
| 0.3125 |
100 |
0.0918 |
- |
| 0.4688 |
150 |
0.0042 |
- |
| 0.625 |
200 |
0.0005 |
- |
| 0.7812 |
250 |
0.0012 |
- |
| 0.9375 |
300 |
0.0005 |
- |
| 1.0938 |
350 |
0.0005 |
- |
| 1.25 |
400 |
0.0003 |
- |
| 1.4062 |
450 |
0.0002 |
- |
| 1.5625 |
500 |
0.0002 |
- |
| 1.7188 |
550 |
0.0001 |
- |
| 1.875 |
600 |
0.0001 |
- |
| 2.0312 |
650 |
0.0002 |
- |
| 2.1875 |
700 |
0.0001 |
- |
| 2.3438 |
750 |
0.0001 |
- |
| 2.5 |
800 |
0.0001 |
- |
| 2.6562 |
850 |
0.0001 |
- |
| 2.8125 |
900 |
0.0001 |
- |
| 2.9688 |
950 |
0.0001 |
- |
| 3.125 |
1000 |
0.0002 |
- |
| 3.2812 |
1050 |
0.0001 |
- |
| 3.4375 |
1100 |
0.0001 |
- |
| 3.5938 |
1150 |
0.0001 |
- |
| 3.75 |
1200 |
0.0001 |
- |
| 3.9062 |
1250 |
0.0001 |
- |
Framework Versions
- Python: 3.10.12
- SetFit: 1.0.2
- Sentence Transformers: 2.2.2
- Transformers: 4.35.2
- PyTorch: 2.1.0+cu121
- Datasets: 2.16.1
- Tokenizers: 0.15.0
Citation
BibTeX
@article{https://doi.org/10.48550/arxiv.2209.11055,
doi = {10.48550/ARXIV.2209.11055},
url = {https://arxiv.org/abs/2209.11055},
author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {Efficient Few-Shot Learning Without Prompts},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}
