Efficient Few-Shot Learning Without Prompts
Paper
•
2209.11055
•
Published
•
4
This is a SetFit model that can be used for Text Classification. This SetFit model uses mini1013/master_domain 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:
| Label | Examples |
|---|---|
| 8.0 |
|
| 6.0 |
|
| 4.0 |
|
| 2.0 |
|
| 0.0 |
|
| 1.0 |
|
| 7.0 |
|
| 3.0 |
|
| 9.0 |
|
| 5.0 |
|
| Label | Accuracy |
|---|---|
| all | 1.0 |
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("mini1013/master_cate_sl12")
# Run inference
preds = model("요넥스 나노지 배드민턴스트링 NBG 98-2 200M 스포츠/레저>배드민턴>스트링")
| Training set | Min | Median | Max |
|---|---|---|---|
| Word count | 2 | 8.0186 | 22 |
| Label | Training Sample Count |
|---|---|
| 0.0 | 70 |
| 1.0 | 70 |
| 2.0 | 70 |
| 3.0 | 16 |
| 4.0 | 70 |
| 5.0 | 70 |
| 6.0 | 70 |
| 7.0 | 70 |
| 8.0 | 70 |
| 9.0 | 69 |
| Epoch | Step | Training Loss | Validation Loss |
|---|---|---|---|
| 0.0079 | 1 | 0.475 | - |
| 0.3968 | 50 | 0.4972 | - |
| 0.7937 | 100 | 0.2864 | - |
| 1.1905 | 150 | 0.1285 | - |
| 1.5873 | 200 | 0.0559 | - |
| 1.9841 | 250 | 0.0233 | - |
| 2.3810 | 300 | 0.007 | - |
| 2.7778 | 350 | 0.0026 | - |
| 3.1746 | 400 | 0.0006 | - |
| 3.5714 | 450 | 0.0004 | - |
| 3.9683 | 500 | 0.0002 | - |
| 4.3651 | 550 | 0.0001 | - |
| 4.7619 | 600 | 0.0001 | - |
| 5.1587 | 650 | 0.0001 | - |
| 5.5556 | 700 | 0.0001 | - |
| 5.9524 | 750 | 0.0001 | - |
| 6.3492 | 800 | 0.0001 | - |
| 6.7460 | 850 | 0.0002 | - |
| 7.1429 | 900 | 0.0001 | - |
| 7.5397 | 950 | 0.0001 | - |
| 7.9365 | 1000 | 0.0 | - |
| 8.3333 | 1050 | 0.0 | - |
| 8.7302 | 1100 | 0.0 | - |
| 9.1270 | 1150 | 0.0 | - |
| 9.5238 | 1200 | 0.0 | - |
| 9.9206 | 1250 | 0.0 | - |
| 10.3175 | 1300 | 0.0 | - |
| 10.7143 | 1350 | 0.0 | - |
| 11.1111 | 1400 | 0.0 | - |
| 11.5079 | 1450 | 0.0 | - |
| 11.9048 | 1500 | 0.0 | - |
| 12.3016 | 1550 | 0.0 | - |
| 12.6984 | 1600 | 0.0 | - |
| 13.0952 | 1650 | 0.0 | - |
| 13.4921 | 1700 | 0.0 | - |
| 13.8889 | 1750 | 0.0 | - |
| 14.2857 | 1800 | 0.0 | - |
| 14.6825 | 1850 | 0.0 | - |
| 15.0794 | 1900 | 0.0 | - |
| 15.4762 | 1950 | 0.0 | - |
| 15.8730 | 2000 | 0.0 | - |
| 16.2698 | 2050 | 0.0 | - |
| 16.6667 | 2100 | 0.0 | - |
| 17.0635 | 2150 | 0.0 | - |
| 17.4603 | 2200 | 0.0 | - |
| 17.8571 | 2250 | 0.0 | - |
| 18.2540 | 2300 | 0.0 | - |
| 18.6508 | 2350 | 0.0 | - |
| 19.0476 | 2400 | 0.0 | - |
| 19.4444 | 2450 | 0.0 | - |
| 19.8413 | 2500 | 0.0 | - |
| 20.2381 | 2550 | 0.0 | - |
| 20.6349 | 2600 | 0.0 | - |
| 21.0317 | 2650 | 0.0 | - |
| 21.4286 | 2700 | 0.0 | - |
| 21.8254 | 2750 | 0.0 | - |
| 22.2222 | 2800 | 0.0 | - |
| 22.6190 | 2850 | 0.0 | - |
| 23.0159 | 2900 | 0.0 | - |
| 23.4127 | 2950 | 0.0 | - |
| 23.8095 | 3000 | 0.0 | - |
| 24.2063 | 3050 | 0.0 | - |
| 24.6032 | 3100 | 0.0 | - |
| 25.0 | 3150 | 0.0 | - |
| 25.3968 | 3200 | 0.0 | - |
| 25.7937 | 3250 | 0.0 | - |
| 26.1905 | 3300 | 0.0 | - |
| 26.5873 | 3350 | 0.0 | - |
| 26.9841 | 3400 | 0.0 | - |
| 27.3810 | 3450 | 0.0 | - |
| 27.7778 | 3500 | 0.0 | - |
| 28.1746 | 3550 | 0.0 | - |
| 28.5714 | 3600 | 0.0 | - |
| 28.9683 | 3650 | 0.0 | - |
| 29.3651 | 3700 | 0.0 | - |
| 29.7619 | 3750 | 0.0 | - |
@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}
}