Efficient Few-Shot Learning Without Prompts
Paper • 2209.11055 • Published • 7
How to use mini1013/master_cate_el8 with setfit:
from setfit import SetFitModel
model = SetFitModel.from_pretrained("mini1013/master_cate_el8")How to use mini1013/master_cate_el8 with sentence-transformers:
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("mini1013/master_cate_el8")
sentences = [
"The weather is lovely today.",
"It's so sunny outside!",
"He drove to the stadium."
]
embeddings = model.encode(sentences)
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]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 |
|---|---|
| 3 |
|
| 2 |
|
| 8 |
|
| 5 |
|
| 4 |
|
| 7 |
|
| 0 |
|
| 6 |
|
| 1 |
|
| 9 |
|
| Label | Metric |
|---|---|
| all | 0.8029 |
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_el8")
# Run inference
preds = model("넥시 CAP02 USB HDMI 캡쳐보드 젠더타입 주식회사 디앤에스티")
| Training set | Min | Median | Max |
|---|---|---|---|
| Word count | 3 | 9.3503 | 26 |
| Label | Training Sample Count |
|---|---|
| 0 | 49 |
| 1 | 25 |
| 2 | 50 |
| 3 | 50 |
| 4 | 50 |
| 5 | 50 |
| 6 | 15 |
| 7 | 50 |
| 8 | 50 |
| 9 | 5 |
| Epoch | Step | Training Loss | Validation Loss |
|---|---|---|---|
| 0.0161 | 1 | 0.496 | - |
| 0.8065 | 50 | 0.2401 | - |
| 1.6129 | 100 | 0.0385 | - |
| 2.4194 | 150 | 0.025 | - |
| 3.2258 | 200 | 0.0181 | - |
| 4.0323 | 250 | 0.0004 | - |
| 4.8387 | 300 | 0.0002 | - |
| 5.6452 | 350 | 0.0001 | - |
| 6.4516 | 400 | 0.0002 | - |
| 7.2581 | 450 | 0.0001 | - |
| 8.0645 | 500 | 0.0001 | - |
| 8.8710 | 550 | 0.0001 | - |
| 9.6774 | 600 | 0.0001 | - |
| 10.4839 | 650 | 0.0001 | - |
| 11.2903 | 700 | 0.0001 | - |
| 12.0968 | 750 | 0.0 | - |
| 12.9032 | 800 | 0.0 | - |
| 13.7097 | 850 | 0.0 | - |
| 14.5161 | 900 | 0.0 | - |
| 15.3226 | 950 | 0.0 | - |
| 16.1290 | 1000 | 0.0 | - |
| 16.9355 | 1050 | 0.0 | - |
| 17.7419 | 1100 | 0.0 | - |
| 18.5484 | 1150 | 0.0 | - |
| 19.3548 | 1200 | 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}
}