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

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/all-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:

Fine-tuning a Sentence Transformer with contrastive learning.
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-mpnet-base-v2
Classification head: a LogisticRegression instance
Maximum Sequence Length: 384 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	'peanut butter cookie dough blizzard is ??????????????????????' 'Free Ebay Sniping RT? http://t.co/B231Ul1O1K Lumbar Extender Back Stretcher Excellent Condition!! ?Please Favorite & Share' "'13 M. Chapoutier Crozes Hermitage so much purple violets slate crushed gravel white pepper. Yum #france #wine #DC http://t.co/skvWN38HZ7"
1	'DUST IN THE WIND: @82ndABNDIV paratroopers move to a loading zone during a dust storm in support of Operation Fury: http://t.co/uGesKLCn8M' 'Delhi Government to Provide Free Treatment to Acid Attack Victims in Private Hospitals http://t.co/H6PM1W7elL' 'National Briefing

Evaluation

Metrics

Label	Accuracy
all	0.8099

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("pEpOo/catastrophy")
# Run inference
preds = model("Heat wave warning aa? Ayyo dei. Just when I plan to visit friends after a year.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	2	15.3737	31

Label	Training Sample Count
0	222
1	158

Training Hyperparameters

batch_size: (8, 8)
num_epochs: (1, 1)
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
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0005	1	0.3038	-
0.0263	50	0.1867	-
0.0526	100	0.2578	-
0.0789	150	0.2298	-
0.1053	200	0.1253	-
0.1316	250	0.0446	-
0.1579	300	0.1624	-
0.1842	350	0.0028	-
0.2105	400	0.0059	-
0.2368	450	0.0006	-
0.2632	500	0.0287	-
0.2895	550	0.003	-
0.3158	600	0.0004	-
0.3421	650	0.0014	-
0.3684	700	0.0002	-
0.3947	750	0.0001	-
0.4211	800	0.0002	-
0.4474	850	0.0002	-
0.4737	900	0.0002	-
0.5	950	0.0826	-
0.5263	1000	0.0002	-
0.5526	1050	0.0001	-
0.5789	1100	0.0003	-
0.6053	1150	0.0303	-
0.6316	1200	0.0001	-
0.6579	1250	0.0	-
0.6842	1300	0.0001	-
0.7105	1350	0.0	-
0.7368	1400	0.0001	-
0.7632	1450	0.0002	-
0.7895	1500	0.0434	-
0.8158	1550	0.0001	-
0.8421	1600	0.0	-
0.8684	1650	0.0001	-
0.8947	1700	0.0001	-
0.9211	1750	0.0001	-
0.9474	1800	0.0001	-
0.9737	1850	0.0001	-
1.0	1900	0.0	-

Framework Versions

Python: 3.10.12
SetFit: 1.0.1
Sentence Transformers: 2.2.2
Transformers: 4.35.2
PyTorch: 2.1.0+cu121
Datasets: 2.15.0
Tokenizers: 0.15.0

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}
}

pEpOo
/

catastrophy