kemonito233/intentino

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: 18 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
14	'tengo otro prestamo activo' 'mi historial esta mal' 'tengo credito con otra financiera'
11	'hable mas fuerte' 'se oye muy lejos' 'se corta'
15	'ahorita voy manejando, hablame luego' 'ahorita no puedo atenderte, estoy ocupado' 'voy manejando'
7	'ya fallecio' 'ya no esta con nosotros' 'el ya no vive'
4	'adios, buenas noches' 'bueno, gracias, adios' 'listo, hasta luego'
10	'si, quiero saber' 'si, digame rapido' 'te escucho'
12	'no, joven, muchas gracias' 'no, oiga, gracias' 'no, por ahora paso, gracias'
17	'bueno, diga' 'si' 'si, diga'
3	'si, a ver de que se trata' 'tal vez si' 'esta bien, envialo'
5	'no corresponde ese numero' 'esta llamando al numero equivocado' 'aqui no vive esa persona'
8	'¿me da la direccion de sus oficinas?' 'yo no les di mi telefono' 'yo no le di mis datos a nadie'
0	'soy su hermana' 'esta bajo tratamiento' 'se siente mal'
16	'¿quien me llama?' '¿de que empresa llaman?' '¿quien es?'
1	'habla el senor' 'con ella habla' 'si aqui habla'
6	'un momento por favor' 'deja le hablo' 'permiteme un segundo, no me cuelgues'
2	'¿con quien quiere hablar?' '¿quien busca?' '¿a quien esta buscando?'
9	'no esten chingando' 'es la quinta vez que me marcan hoy' '¡que no entiendes que no!'
13	'salio a la tienda, no tarda' 'ahorita no esta, anda de viaje' 'anda trabajando'

Evaluation

Metrics

Label	Accuracy
all	0.9111

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("setfit_model_id")
# Run inference
preds = model("soy quien busca")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	3.9018	11

Label	Training Sample Count
0	32
1	18
2	11
3	18
4	18
5	22
6	9
7	12
8	40
9	11
10	33
11	13
12	48
13	8
14	36
15	13
16	18
17	37

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 20
• 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: True

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0010	1	0.3888	-
0.0504	50	0.211	-
0.1007	100	0.1344	-
0.1511	150	0.0742	-
0.2014	200	0.0484	-
0.2518	250	0.0387	-
0.3021	300	0.0264	-
0.3525	350	0.0183	-
0.4028	400	0.0135	-
0.4532	450	0.0115	-
0.5035	500	0.0082	-
0.5539	550	0.0083	-
0.6042	600	0.0073	-
0.6546	650	0.009	-
0.7049	700	0.0067	-
0.7553	750	0.0075	-
0.8056	800	0.0085	-
0.8560	850	0.0073	-
0.9063	900	0.0065	-
0.9567	950	0.0076	-
1.0	993	-	0.0437

Framework Versions

• Python: 3.12.12
• SetFit: 1.1.3
• Sentence Transformers: 5.2.2
• Transformers: 4.44.2
• PyTorch: 2.9.0+cu126
• Datasets: 4.0.0
• 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}
}