SetFit

This is a SetFit model that can be used for Text Classification. A LinearSVC 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
Classification head: a LinearSVC 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
1	'Climate change and bad development practices to blame for Kerala floods, says expert: https://t.co/YCMHzycfLO https://t.co/voKu0dqcMQ' "Canada's chief public health officer says no vaccine for coronavirus for a\xa0year https://t.co/itrdTgqSiZ https://t.co/Ohjw1uwwC2" 'The death toll from a new #coronavirus in China rose sharply to 132 on Wednesday with nearly 1,500 new cases, as Japan said it flew citizens out of Wuhan, the epicentre of the virus. More here: https://t.co/v2PITpjrcE https://t.co/HSjwwBXG5H'
0	'Don't Be So Afraid Of The Coronavirus.\nIt Won't Last Long.\nBecause "MADE IN CHINA"\n🤔🤔🤔🤔🤣🤣🤣🤣 https://t.co/4qlRbmMevC' 'Why Are Ecosystems Collapsing Worldwide? Learn more here: https://t.co/fgSyGEvKaA\n#climatechange #climateaction \n#environment \n#energy https://t.co/ac0E1xpWrd' 'What can I do about the Wuhan coronavirus? A guide for Canadians of what’s helpful, and what’s not https://t.co/xwygAwhUJZ https://t.co/GC2iYMWM2l'

Label

Examples

'Climate change and bad development practices to blame for Kerala floods, says expert: https://t.co/YCMHzycfLO https://t.co/voKu0dqcMQ'
"Canada's chief public health officer says no vaccine for coronavirus for a\xa0year https://t.co/itrdTgqSiZ https://t.co/Ohjw1uwwC2"
'The death toll from a new #coronavirus in China rose sharply to 132 on Wednesday with nearly 1,500 new cases, as Japan said it flew citizens out of Wuhan, the epicentre of the virus. More here: https://t.co/v2PITpjrcE https://t.co/HSjwwBXG5H'

'Don't Be So Afraid Of The Coronavirus.\nIt Won't Last Long.\nBecause "MADE IN CHINA"\n🤔🤔🤔🤔🤣🤣🤣🤣 https://t.co/4qlRbmMevC'
'Why Are Ecosystems Collapsing Worldwide? Learn more here: https://t.co/fgSyGEvKaA\n#climatechange #climateaction \n#environment \n#energy https://t.co/ac0E1xpWrd'
'What can I do about the Wuhan coronavirus? A guide for Canadians of what’s helpful, and what’s not https://t.co/xwygAwhUJZ https://t.co/GC2iYMWM2l'

Evaluation

Metrics

Label	Accuracy
all	0.8164

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("SOUMYADEEPSAR/SetFit_Clef_task1")
# Run inference
preds = model("Easy trade set up on #bitcoin

Inside bar on the daily. Long the break out, short the breakdown. https://t.co/rzfdY37ZDd")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	4	20.955	53

Label	Training Sample Count
0	100
1	100

Training Hyperparameters

batch_size: (32, 32)
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
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0016	1	0.2832	-
0.0791	50	0.2626	-
0.1582	100	0.2525	-
0.2373	150	0.1303	-
0.3165	200	0.0029	-
0.3956	250	0.0019	-
0.4747	300	0.0014	-
0.5538	350	0.001	-
0.6329	400	0.001	-
0.7120	450	0.0008	-
0.7911	500	0.0008	-
0.8703	550	0.0007	-
0.9494	600	0.0006	-
1.0285	650	0.0007	-
1.1076	700	0.0006	-
1.1867	750	0.0006	-
1.2658	800	0.0005	-
1.3449	850	0.0005	-
1.4241	900	0.0005	-
1.5032	950	0.0005	-
1.5823	1000	0.0005	-
1.6614	1050	0.0005	-
1.7405	1100	0.0005	-
1.8196	1150	0.0005	-
1.8987	1200	0.0004	-
1.9778	1250	0.0004	-
2.0570	1300	0.0004	-
2.1361	1350	0.0005	-
2.2152	1400	0.0004	-
2.2943	1450	0.0004	-
2.3734	1500	0.0004	-
2.4525	1550	0.0006	-
2.5316	1600	0.0004	-
2.6108	1650	0.0003	-
2.6899	1700	0.0004	-
2.7690	1750	0.0004	-
2.8481	1800	0.0004	-
2.9272	1850	0.0004	-

Framework Versions

Python: 3.10.12
SetFit: 1.0.3
Sentence Transformers: 2.3.1
Transformers: 4.37.2
PyTorch: 2.1.0+cu121
Datasets: 2.17.1
Tokenizers: 0.15.2

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}
}

SOUMYADEEPSAR
/

SetFit_Clef_task1