adriansanz/gret4

SetFit with projecte-aina/ST-NLI-ca_paraphrase-multilingual-mpnet-base

This is a SetFit model that can be used for Text Classification. This SetFit model uses projecte-aina/ST-NLI-ca_paraphrase-multilingual-mpnet-base 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: projecte-aina/ST-NLI-ca_paraphrase-multilingual-mpnet-base
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 128 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
0	"Sóc ciutadà i m'agradaria saber quin és el tràmit per a la renovació del DNI." "Quin és el propòsit de la garantia per a l'abocament controlat de runes?" 'Quin és el benefici de la devolució de fiances i avals?'
1	"Aquest text és Saludo per a un cercador de tràmits d'un ajuntament" 'Bon dia, vull saber més sobre els tràmits disponibles.' "Bona nit, com t'has anat acostant al final del dia?"

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("adriansanz/gret4")
# Run inference
preds = model("Hola!")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	9.3444	17

Label	Training Sample Count
0	45
1	45

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (3, 3)
• 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
• evaluation_strategy: epoch
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0039	1	0.2366	-
0.1931	50	0.1287	-
0.3861	100	0.0039	-
0.5792	150	0.0003	-
0.7722	200	0.0001	-
0.9653	250	0.0001	-
1.0	259	-	0.0001
1.1583	300	0.0001	-
1.3514	350	0.0001	-
1.5444	400	0.0001	-
1.7375	450	0.0001	-
1.9305	500	0.0001	-
2.0	518	-	0.0001
2.1236	550	0.0	-
2.3166	600	0.0	-
2.5097	650	0.0	-
2.7027	700	0.0	-
2.8958	750	0.0	-
3.0	777	-	0.0001

Framework Versions

• Python: 3.10.12
• SetFit: 1.1.0
• Sentence Transformers: 3.2.1
• Transformers: 4.42.2
• PyTorch: 2.5.0+cu121
• Datasets: 3.1.0
• 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}
}