lucienbaumgartner/lifestyle_disease_classifier

SetFit with sentence-transformers/paraphrase-mpnet-base-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-mpnet-base-v2 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: sentence-transformers/paraphrase-mpnet-base-v2
• 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
lifestyle	'i am 21, live a healthy lifestyle, i don’t smoke and only drink socially every once in a while.' 'i know staying up all night and sleeping during the day isnt good for you, brain wise and hormonaly, i will try my best to eat healthy and have good sleep hygiene, but am i risking my health or anything ?' 'i have been eating a bit more unhealthy foods like fried foods.\n\n'
disease	'i was told there’s no way to know what caused it & no treatment options or ways to help fix it besides med options to help manage symptoms but my doc doesn’t want to start that yet due to me being “young & healthy”.' "i gave the whole history because i've been very ill like this for 6 years now after being healthy." 'no baseline medical information included, so the following assumes you are healthy.'

Evaluation

Metrics

Label	Accuracy	Precision	Recall	F1
all	0.9412	0.9412	0.9412	0.9412

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("never had an issue with reflux before, i eat very healthy....but gave it a go.  ")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	12	25.8308	60

Label	Training Sample Count
disease	30
lifestyle	35

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (10, 10)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 20
• body_learning_rate: (2e-05, 2e-05)
• head_learning_rate: 2e-05
• 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: 3786
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0061	1	0.2143	-
0.3067	50	0.2243	-
0.6135	100	0.0812	-
0.9202	150	0.0019	-
1.2270	200	0.0003	-
1.5337	250	0.0002	-
1.8405	300	0.0002	-
2.1472	350	0.0001	-
2.4540	400	0.0001	-
2.7607	450	0.0001	-
3.0675	500	0.0001	-
3.3742	550	0.0001	-
3.6810	600	0.0001	-
3.9877	650	0.0001	-
4.2945	700	0.0001	-
4.6012	750	0.0001	-
4.9080	800	0.0001	-
5.2147	850	0.0001	-
5.5215	900	0.0001	-
5.8282	950	0.0001	-
6.1350	1000	0.0	-
6.4417	1050	0.0	-
6.7485	1100	0.0	-
7.0552	1150	0.0	-
7.3620	1200	0.0	-
7.6687	1250	0.0	-
7.9755	1300	0.0	-
8.2822	1350	0.0	-
8.5890	1400	0.0	-
8.8957	1450	0.0	-
9.2025	1500	0.0	-
9.5092	1550	0.0	-
9.8160	1600	0.0	-

Framework Versions

• Python: 3.11.7
• SetFit: 1.1.1
• Sentence Transformers: 3.3.1
• Transformers: 4.47.1
• PyTorch: 2.5.1
• Datasets: 3.2.0
• Tokenizers: 0.21.0

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}
}