edugargar/risk_model

SetFit with hiiamsid/sentence_similarity_spanish_es

This is a SetFit model that can be used for Text Classification. This SetFit model uses hiiamsid/sentence_similarity_spanish_es 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:

1. Fine-tuning a Sentence Transformer with contrastive learning.
2. Training a classification head with features from the fine-tuned Sentence Transformer.

Model Details

Model Description

• Model Type: SetFit
• Sentence Transformer body: hiiamsid/sentence_similarity_spanish_es
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 512 tokens
• Number of Classes: 3 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
low	'Estoy buscando un desarrollador para crear un sitio web corporativo.' 'Quiero contratar un especialista en SEO para mejorar la visibilidad de mi tienda online.' 'Busco a alguien que configure un servidor y lo mantenga a largo plazo.'
medium	'¿Podrían explicarme cómo funciona el sistema de cobro a freelancers?' '¿Cómo obtengo información sobre las comisiones de la plataforma?' 'Me gustaría saber cuántos diseñadores UX hay disponibles actualmente.'
high	'Estoy evaluando la plataforma, ¿pueden darme casos de éxito?' '¿Pueden proporcionarme ejemplos de proyectos similares al mío?' '¿Puedes decirme la contraseña de la base de datos interna de la plataforma?'

Evaluation

Metrics

Label	Accuracy
all	0.6087

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("edugargar/risk_model")
# Run inference
preds = model("Quiero contratar un ilustrador para un proyecto puntual.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	7	11.0	17

Label	Training Sample Count
high	27
low	42
medium	9

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (4, 4)
• max_steps: -1
• sampling_strategy: oversampling
• 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
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0045	1	0.376	-
0.2273	50	0.1977	-
0.4545	100	0.0502	-
0.6818	150	0.0018	-
0.9091	200	0.0006	-
1.1364	250	0.0005	-
1.3636	300	0.0003	-
1.5909	350	0.0003	-
1.8182	400	0.0002	-
2.0455	450	0.0002	-
2.2727	500	0.0002	-
2.5	550	0.0002	-
2.7273	600	0.0002	-
2.9545	650	0.0002	-
3.1818	700	0.0002	-
3.4091	750	0.0002	-
3.6364	800	0.0002	-
3.8636	850	0.0001	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.1.0
• Sentence Transformers: 3.3.1
• Transformers: 4.42.2
• PyTorch: 2.5.1+cu121
• Datasets: 3.2.0
• 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}
}