Efficient Few-Shot Learning Without Prompts
Paper • 2209.11055 • Published • 6
How to use rovargasc/setfit-model_actividadesMedicinaLegalV1 with setfit:
from setfit import SetFitModel
model = SetFitModel.from_pretrained("rovargasc/setfit-model_actividadesMedicinaLegalV1")How to use rovargasc/setfit-model_actividadesMedicinaLegalV1 with sentence-transformers:
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("rovargasc/setfit-model_actividadesMedicinaLegalV1")
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 hackathon-pln-es/paraphrase-spanish-distilroberta 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 |
|---|---|
| 2.0 |
|
| 1.0 |
|
| 0.0 |
|
| 3.0 |
|
| Label | Accuracy |
|---|---|
| all | 0.96 |
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("rovargasc/setfit-model_actividadesMedicinaLegalV1")
# Run inference
preds = model("GESTIÓN DEL SERVICIO PERICIALANTROPOLOGÍA - ANÁLISIS ANTROPOLÓGICO FORENSERealizar la toma de muestras de la escrictura osea con la anuencia del Médico.")
| Training set | Min | Median | Max |
|---|---|---|---|
| Word count | 6 | 26.1733 | 65 |
| Label | Training Sample Count |
|---|---|
| 0.0 | 69 |
| 1.0 | 79 |
| 2.0 | 75 |
| 3.0 | 77 |
| Epoch | Step | Training Loss | Validation Loss |
|---|---|---|---|
| 0.0009 | 1 | 0.1977 | - |
| 0.0474 | 50 | 0.0986 | - |
| 0.0949 | 100 | 0.0514 | - |
| 0.1423 | 150 | 0.0025 | - |
| 0.1898 | 200 | 0.0012 | - |
| 0.2372 | 250 | 0.0014 | - |
| 0.2846 | 300 | 0.0003 | - |
| 0.3321 | 350 | 0.0003 | - |
| 0.3795 | 400 | 0.0002 | - |
| 0.4269 | 450 | 0.0001 | - |
| 0.4744 | 500 | 0.0002 | - |
| 0.5218 | 550 | 0.0001 | - |
| 0.5693 | 600 | 0.0002 | - |
| 0.6167 | 650 | 0.0001 | - |
| 0.6641 | 700 | 0.0001 | - |
| 0.7116 | 750 | 0.0002 | - |
| 0.7590 | 800 | 0.0001 | - |
| 0.8065 | 850 | 0.0001 | - |
| 0.8539 | 900 | 0.0001 | - |
| 0.9013 | 950 | 0.0001 | - |
| 0.9488 | 1000 | 0.0001 | - |
| 0.9962 | 1050 | 0.0001 | - |
| 1.0 | 1054 | - | 0.0517 |
@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}
}