Corran/SciGenSetfit24Binary

SetFit with sentence-transformers/all-MiniLM-L6-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/all-MiniLM-L6-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/all-MiniLM-L6-v2
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 256 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
Misc	'Pravastatin therapy in patients with average cholesterol levels following myocardial infarction has been shown to reduce the risk of coronary events, highlighting the importance of lipid-lowering therapy in internal medicine for cardiovascular disease prevention.' 'However, the efficacy of pravastatin in patients with average cholesterol levels is less clear.' 'This study investigates the impact of Pravastatin on reducing coronary events in internal medicine patients with average cholesterol levels after a myocardial infarction.'
Uncertainty	'Despite the widespread use of pravastatin in post-myocardial infarction patients with average cholesterol levels, the evidence regarding its impact on coronary events remains inconclusive and sometimes contradictory.' 'Despite the findings of this study showing a reduction in coronary events with Pravastatin use in patients with average cholesterol levels, contrasting evidence exists suggesting no significant benefit in similar patient populations (Miller et al., 2018).' 'Despite the proven benefits of dual antiplatelet therapy with aspirin and clopidogrel in the secondary prevention of cardiovascular events, particularly in coronary artery disease, there is a paucity of data specifically addressing its use in stroke or transient ischemic attack (TIA) patients.'

Evaluation

Metrics

Label	Accuracy
all	0.9498

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("Corran/SciGenSetfit24Binary")
# Run inference
preds = model("The study reported that 73% of the psychotherapists endorsed the use of cognitive techniques in their treatment of eating disorders, while 61% reported using behavioral techniques.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	8	29.6038	60

Label	Training Sample Count
Misc	2500
Uncertainty	2500

Training Hyperparameters

• batch_size: (300, 300)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 5
• 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
0.0060	1	0.4529	-
0.2994	50	0.3104	-
0.5988	100	0.2514	-
0.8982	150	0.25	-
1.0	167	-	0.2479
0.0060	1	0.2406	-
0.2994	50	0.1576	-
0.5988	100	0.0912	-
0.8982	150	0.0656	-
1.0	167	-	0.0683
0.0060	1	0.0827	-
0.2994	50	0.0581	-
0.5988	100	0.0393	-
0.8982	150	0.0339	-
1.0	167	-	0.0516

Framework Versions

• Python: 3.10.12
• SetFit: 1.2.0.dev0
• Sentence Transformers: 3.1.1
• 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}
}