SetFit
This is a SetFit model that can be used for Text Classification. 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 Type: SetFit
- Classification head: a LogisticRegression instance
- Maximum Sequence Length: 512 tokens
- Number of Classes: 20 classes
Model Sources
- Repository: SetFit on GitHub
- Paper: Efficient Few-Shot Learning Without Prompts
- Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts
Model Labels
Label | Examples |
---|---|
2014 |
|
2001 |
|
2026 |
|
2013 |
|
1001 |
|
304 |
|
237 |
|
2038 |
|
49 |
|
357 |
|
2022 |
|
2017 |
|
78 |
|
2037 |
|
2039 |
|
353 |
|
2002 |
|
2010 |
|
994 |
|
2060 |
|
Evaluation
Metrics
Label | Accuracy |
---|---|
all | 0.8 |
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
# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("desarrolloasesoreslocales/bert-leg-al-setfit-grok")
# Run inference
preds = model("procedo al estacionamiento por autorización del agente 12289 (Policia laca. Aa vuelta en 5 minutos, me encuentro con una multa LL5898790 por parte del agente 12312")
Training Details
Training Set Metrics
Training set | Min | Median | Max |
---|---|---|---|
Word count | 4 | 44.1625 | 212 |
Label | Training Sample Count |
---|---|
49 | 8 |
78 | 8 |
237 | 8 |
304 | 8 |
353 | 8 |
357 | 8 |
994 | 8 |
1001 | 8 |
2001 | 8 |
2002 | 8 |
2010 | 8 |
2013 | 8 |
2014 | 8 |
2017 | 8 |
2022 | 8 |
2026 | 8 |
2037 | 8 |
2038 | 8 |
2039 | 8 |
2060 | 8 |
Training Hyperparameters
- batch_size: (16, 16)
- num_epochs: (10, 10)
- max_steps: -1
- sampling_strategy: oversampling
- body_learning_rate: (1e-06, 1e-06)
- head_learning_rate: 0.003
- loss: CosineSimilarityLoss
- distance_metric: cosine_distance
- margin: 0.25
- end_to_end: True
- use_amp: True
- warmup_proportion: 0.1
- l2_weight: 0.001
- seed: 42
- eval_max_steps: 100
- load_best_model_at_end: True
Training Results
Epoch | Step | Training Loss | Validation Loss |
---|---|---|---|
0.0007 | 1 | 0.0031 | - |
0.0658 | 100 | 0.0003 | 0.067 |
0.1316 | 200 | 0.0001 | 0.0717 |
0.1974 | 300 | 0.0001 | 0.0711 |
0.2632 | 400 | 0.0003 | 0.0721 |
0.3289 | 500 | 0.0021 | 0.0667 |
0.3947 | 600 | 0.0001 | 0.0611 |
0.4605 | 700 | 0.0002 | 0.0672 |
0.5263 | 800 | 0.0001 | 0.0777 |
0.5921 | 900 | 0.0001 | 0.067 |
0.6579 | 1000 | 0.0001 | 0.0687 |
0.7237 | 1100 | 0.0 | 0.0661 |
0.7895 | 1200 | 0.005 | 0.0695 |
0.8553 | 1300 | 0.0004 | 0.0661 |
0.9211 | 1400 | 0.0019 | 0.0667 |
0.9868 | 1500 | 0.0001 | 0.0672 |
1.0526 | 1600 | 0.0001 | 0.0714 |
1.1184 | 1700 | 0.0001 | 0.0687 |
1.1842 | 1800 | 0.0001 | 0.0723 |
1.25 | 1900 | 0.0 | 0.0722 |
1.3158 | 2000 | 0.0001 | 0.0728 |
1.3816 | 2100 | 0.0 | 0.0713 |
1.4474 | 2200 | 0.0 | 0.0733 |
1.5132 | 2300 | 0.0025 | 0.0719 |
1.5789 | 2400 | 0.0 | 0.0708 |
1.6447 | 2500 | 0.0 | 0.0722 |
1.7105 | 2600 | 0.0 | 0.0723 |
1.7763 | 2700 | 0.0 | 0.069 |
1.8421 | 2800 | 0.0 | 0.0703 |
1.9079 | 2900 | 0.0 | 0.0722 |
1.9737 | 3000 | 0.0001 | 0.0701 |
2.0395 | 3100 | 0.0 | 0.0691 |
2.1053 | 3200 | 0.0024 | 0.0706 |
2.1711 | 3300 | 0.0001 | 0.0716 |
2.2368 | 3400 | 0.0001 | 0.0886 |
2.3026 | 3500 | 0.0011 | 0.0734 |
2.3684 | 3600 | 0.0001 | 0.0875 |
2.4342 | 3700 | 0.0001 | 0.0809 |
2.5 | 3800 | 0.0 | 0.0818 |
2.5658 | 3900 | 0.0001 | 0.0829 |
2.6316 | 4000 | 0.0 | 0.0833 |
2.6974 | 4100 | 0.0036 | 0.0841 |
2.7632 | 4200 | 0.0 | 0.0833 |
2.8289 | 4300 | 0.0 | 0.0831 |
2.8947 | 4400 | 0.0374 | 0.083 |
2.9605 | 4500 | 0.0 | 0.083 |
3.0263 | 4600 | 0.0001 | 0.0831 |
3.0921 | 4700 | 0.0 | 0.0829 |
3.1579 | 4800 | 0.0 | 0.0828 |
3.2237 | 4900 | 0.0 | 0.0828 |
3.2895 | 5000 | 0.0068 | 0.0829 |
3.3553 | 5100 | 0.0 | 0.0826 |
3.4211 | 5200 | 0.0 | 0.0827 |
3.4868 | 5300 | 0.0 | 0.0824 |
3.5526 | 5400 | 0.0 | 0.0823 |
3.6184 | 5500 | 0.0 | 0.0822 |
3.6842 | 5600 | 0.0 | 0.0821 |
3.75 | 5700 | 0.0 | 0.0822 |
3.8158 | 5800 | 0.0 | 0.082 |
3.8816 | 5900 | 0.0032 | 0.0819 |
3.9474 | 6000 | 0.0 | 0.0822 |
4.0132 | 6100 | 0.0 | 0.0824 |
4.0789 | 6200 | 0.0 | 0.0822 |
4.1447 | 6300 | 0.0 | 0.0819 |
4.2105 | 6400 | 0.0 | 0.0822 |
4.2763 | 6500 | 0.0057 | 0.0824 |
4.3421 | 6600 | 0.0 | 0.0824 |
4.4079 | 6700 | 0.0 | 0.0824 |
4.4737 | 6800 | 0.0022 | 0.0822 |
4.5395 | 6900 | 0.0 | 0.0822 |
4.6053 | 7000 | 0.0 | 0.0823 |
4.6711 | 7100 | 0.0 | 0.0822 |
4.7368 | 7200 | 0.0034 | 0.0822 |
4.8026 | 7300 | 0.0 | 0.0822 |
4.8684 | 7400 | 0.0 | 0.0822 |
4.9342 | 7500 | 0.0 | 0.0822 |
5.0 | 7600 | 0.0 | 0.0822 |
0.0007 | 1 | 0.0018 | - |
0.0658 | 100 | 0.0002 | 0.0612 |
0.1316 | 200 | 0.0002 | 0.0613 |
0.1974 | 300 | 0.0002 | 0.0615 |
0.2632 | 400 | 0.0 | 0.0619 |
0.3289 | 500 | 0.0021 | 0.0626 |
0.3947 | 600 | 0.0001 | 0.0628 |
0.4605 | 700 | 0.0001 | 0.0633 |
0.5263 | 800 | 0.0001 | 0.064 |
0.5921 | 900 | 0.0001 | 0.0635 |
0.6579 | 1000 | 0.0001 | 0.0645 |
0.7237 | 1100 | 0.0001 | 0.0659 |
0.7895 | 1200 | 0.0055 | 0.0662 |
0.8553 | 1300 | 0.0001 | 0.0667 |
0.9211 | 1400 | 0.0032 | 0.0673 |
0.9868 | 1500 | 0.0001 | 0.067 |
1.0526 | 1600 | 0.0001 | 0.0668 |
1.1184 | 1700 | 0.0001 | 0.0667 |
1.1842 | 1800 | 0.0001 | 0.0664 |
1.25 | 1900 | 0.0001 | 0.0667 |
1.3158 | 2000 | 0.0 | 0.0674 |
1.3816 | 2100 | 0.0001 | 0.0667 |
1.4474 | 2200 | 0.0 | 0.0669 |
1.5132 | 2300 | 0.0028 | 0.0669 |
1.5789 | 2400 | 0.0001 | 0.0671 |
1.6447 | 2500 | 0.0001 | 0.0676 |
1.7105 | 2600 | 0.0001 | 0.0689 |
1.7763 | 2700 | 0.0001 | 0.069 |
1.8421 | 2800 | 0.0001 | 0.0691 |
1.9079 | 2900 | 0.0001 | 0.0696 |
1.9737 | 3000 | 0.0001 | 0.0688 |
2.0395 | 3100 | 0.0 | 0.0678 |
2.1053 | 3200 | 0.0027 | 0.0677 |
2.1711 | 3300 | 0.0001 | 0.0675 |
2.2368 | 3400 | 0.0 | 0.0676 |
2.3026 | 3500 | 0.0001 | 0.068 |
2.3684 | 3600 | 0.0001 | 0.0672 |
2.4342 | 3700 | 0.0 | 0.0669 |
2.5 | 3800 | 0.0 | 0.0667 |
2.5658 | 3900 | 0.0 | 0.0673 |
2.6316 | 4000 | 0.0 | 0.0672 |
2.6974 | 4100 | 0.0032 | 0.0689 |
2.7632 | 4200 | 0.0 | 0.0691 |
2.8289 | 4300 | 0.0001 | 0.0693 |
2.8947 | 4400 | 0.0388 | 0.0692 |
2.9605 | 4500 | 0.0001 | 0.0691 |
3.0263 | 4600 | 0.0 | 0.0683 |
3.0921 | 4700 | 0.0 | 0.0685 |
3.1579 | 4800 | 0.0001 | 0.0681 |
3.2237 | 4900 | 0.0 | 0.0677 |
3.2895 | 5000 | 0.0081 | 0.0684 |
3.3553 | 5100 | 0.0 | 0.0685 |
3.4211 | 5200 | 0.0 | 0.0681 |
3.4868 | 5300 | 0.0001 | 0.0683 |
3.5526 | 5400 | 0.0001 | 0.0681 |
3.6184 | 5500 | 0.0 | 0.0675 |
3.6842 | 5600 | 0.0 | 0.0687 |
3.75 | 5700 | 0.0001 | 0.0692 |
3.8158 | 5800 | 0.0 | 0.0695 |
3.8816 | 5900 | 0.0038 | 0.069 |
3.9474 | 6000 | 0.0001 | 0.069 |
4.0132 | 6100 | 0.0 | 0.0684 |
4.0789 | 6200 | 0.0001 | 0.0688 |
4.1447 | 6300 | 0.0 | 0.0682 |
4.2105 | 6400 | 0.0 | 0.0677 |
4.2763 | 6500 | 0.0049 | 0.0678 |
4.3421 | 6600 | 0.0001 | 0.068 |
4.4079 | 6700 | 0.0 | 0.0679 |
4.4737 | 6800 | 0.0029 | 0.0679 |
4.5395 | 6900 | 0.0 | 0.0684 |
4.6053 | 7000 | 0.0 | 0.0678 |
4.6711 | 7100 | 0.0 | 0.0688 |
4.7368 | 7200 | 0.004 | 0.0695 |
4.8026 | 7300 | 0.0 | 0.0696 |
4.8684 | 7400 | 0.0 | 0.0695 |
4.9342 | 7500 | 0.0 | 0.0695 |
5.0 | 7600 | 0.0 | 0.0691 |
5.0658 | 7700 | 0.0033 | 0.0691 |
5.1316 | 7800 | 0.0 | 0.0691 |
5.1974 | 7900 | 0.0 | 0.0688 |
5.2632 | 8000 | 0.0 | 0.0689 |
5.3289 | 8100 | 0.0001 | 0.0689 |
5.3947 | 8200 | 0.0 | 0.0688 |
5.4605 | 8300 | 0.0 | 0.0685 |
5.5263 | 8400 | 0.0 | 0.0688 |
5.5921 | 8500 | 0.0 | 0.0683 |
5.6579 | 8600 | 0.003 | 0.0688 |
5.7237 | 8700 | 0.0 | 0.0698 |
5.7895 | 8800 | 0.0037 | 0.0701 |
5.8553 | 8900 | 0.0 | 0.0701 |
5.9211 | 9000 | 0.0001 | 0.0695 |
5.9868 | 9100 | 0.0001 | 0.0697 |
6.0526 | 9200 | 0.0 | 0.0694 |
6.1184 | 9300 | 0.0 | 0.0689 |
6.1842 | 9400 | 0.0 | 0.0686 |
6.25 | 9500 | 0.0025 | 0.0686 |
6.3158 | 9600 | 0.0 | 0.069 |
6.3816 | 9700 | 0.0 | 0.069 |
6.4474 | 9800 | 0.0 | 0.0687 |
6.5132 | 9900 | 0.0001 | 0.0683 |
6.5789 | 10000 | 0.0 | 0.0684 |
6.6447 | 10100 | 0.0 | 0.0684 |
6.7105 | 10200 | 0.0001 | 0.069 |
6.7763 | 10300 | 0.0 | 0.0694 |
6.8421 | 10400 | 0.0028 | 0.0696 |
6.9079 | 10500 | 0.0 | 0.0697 |
6.9737 | 10600 | 0.0 | 0.0697 |
7.0395 | 10700 | 0.0 | 0.0694 |
7.1053 | 10800 | 0.0 | 0.0692 |
7.1711 | 10900 | 0.0 | 0.069 |
7.2368 | 11000 | 0.0 | 0.0691 |
7.3026 | 11100 | 0.0 | 0.0691 |
7.3684 | 11200 | 0.0 | 0.0691 |
7.4342 | 11300 | 0.0025 | 0.069 |
7.5 | 11400 | 0.0 | 0.0687 |
7.5658 | 11500 | 0.0 | 0.0688 |
7.6316 | 11600 | 0.0 | 0.0688 |
7.6974 | 11700 | 0.0001 | 0.0691 |
7.7632 | 11800 | 0.0 | 0.0692 |
7.8289 | 11900 | 0.0001 | 0.0692 |
7.8947 | 12000 | 0.0405 | 0.0693 |
7.9605 | 12100 | 0.0 | 0.0695 |
8.0263 | 12200 | 0.0029 | 0.0694 |
8.0921 | 12300 | 0.0001 | 0.0693 |
8.1579 | 12400 | 0.0 | 0.0692 |
8.2237 | 12500 | 0.0001 | 0.0691 |
8.2895 | 12600 | 0.0045 | 0.0693 |
8.3553 | 12700 | 0.0 | 0.0693 |
8.4211 | 12800 | 0.0 | 0.0692 |
8.4868 | 12900 | 0.0 | 0.0691 |
8.5526 | 13000 | 0.0 | 0.0691 |
8.6184 | 13100 | 0.0026 | 0.069 |
8.6842 | 13200 | 0.0 | 0.0692 |
8.75 | 13300 | 0.0 | 0.0694 |
8.8158 | 13400 | 0.0 | 0.0694 |
8.8816 | 13500 | 0.0 | 0.0693 |
8.9474 | 13600 | 0.0 | 0.0694 |
9.0132 | 13700 | 0.0 | 0.0693 |
9.0789 | 13800 | 0.0 | 0.0693 |
9.1447 | 13900 | 0.0 | 0.0692 |
9.2105 | 14000 | 0.003 | 0.0692 |
9.2763 | 14100 | 0.0044 | 0.0692 |
9.3421 | 14200 | 0.0 | 0.0692 |
9.4079 | 14300 | 0.0 | 0.0692 |
9.4737 | 14400 | 0.0 | 0.0691 |
9.5395 | 14500 | 0.0 | 0.0691 |
9.6053 | 14600 | 0.0 | 0.0691 |
9.6711 | 14700 | 0.0 | 0.0691 |
9.7368 | 14800 | 0.0043 | 0.0692 |
9.8026 | 14900 | 0.0028 | 0.0692 |
9.8684 | 15000 | 0.0 | 0.0692 |
9.9342 | 15100 | 0.0 | 0.0692 |
10.0 | 15200 | 0.0 | 0.0692 |
- The bold row denotes the saved checkpoint.
Framework Versions
- Python: 3.10.12
- SetFit: 1.0.3
- Sentence Transformers: 3.0.1
- Transformers: 4.40.2
- PyTorch: 2.3.0+cu121
- Datasets: 2.20.0
- Tokenizers: 0.19.1
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}
}
- Downloads last month
- 4
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.