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: 2 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
Evento	'El dólar vuelve a subir a la espera de una decisión clave del Banco Central' 'Viernes caluroso y sin lluvias' 'ARA San Juan
Perspectiva	'El futuro de la educación tras la pandemia: ¿hacia un modelo híbrido permanente?' '¿Cómo impacta la automatización en los trabajos de baja calificación?' 'Feminicidios: Falta una construcción social y cultural contra la violencia'

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("EmanuelOrler/setfit-spanish-event-perspective")
# Run inference
preds = model("Hallaron una inmensa `huella digital` en el espacio")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	5	12.9231	24

Label	Training Sample Count
Evento	22
Perspectiva	17

Training Hyperparameters

batch_size: (12, 12)
num_epochs: (4, 16)
max_steps: -1
sampling_strategy: undersampling
body_learning_rate: (2e-05, 1e-05)
head_learning_rate: 0.01
loss: CosineSimilarityLoss
distance_metric: cosine_distance
margin: 0.25
end_to_end: False
use_amp: False
warmup_proportion: 0.1
l2_weight: 0.01
seed: 42
evaluation_strategy: steps
eval_max_steps: -1
load_best_model_at_end: True

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0159	1	0.0885	-
0.1587	10	0.3927	0.2944
0.3175	20	0.3039	0.2387
0.4762	30	0.2466	0.1807
0.6349	40	0.2049	0.1686
0.7937	50	0.1803	0.1786
0.9524	60	0.1319	0.2002
1.1111	70	0.045	0.3103
1.2698	80	0.0099	0.3200
1.4286	90	0.0036	0.3845
1.5873	100	0.0021	0.4078
1.7460	110	0.0011	0.4184
1.9048	120	0.0011	0.4186
2.0635	130	0.0009	0.4282
2.2222	140	0.0008	0.4242
2.3810	150	0.0008	0.4269
2.5397	160	0.0007	0.4303
2.6984	170	0.0006	0.4301
2.8571	180	0.0006	0.4321
3.0159	190	0.0006	0.4311
3.1746	200	0.0005	0.4291
3.3333	210	0.0006	0.4322
3.4921	220	0.0005	0.4315
3.6508	230	0.0005	0.4308
3.8095	240	0.0005	0.4307
3.9683	250	0.0004	0.4312

Framework Versions

Python: 3.10.14
SetFit: 1.1.0
Sentence Transformers: 3.2.1
Transformers: 4.44.0
PyTorch: 2.4.0
Datasets: 2.21.0
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}
}