SOUMYADEEPSAR/Setfit_random_sample_svm_head

SetFit

This is a SetFit model that can be used for Text Classification. A SVC 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
• Classification head: a SVC 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	'ESG funds often charge many times more for investment funds that are nearly indistinguishable from those without the ESG title.' 'They are California, Florida, Illinois, Nebraska, New York, and Wyoming.' 'And so it goes.'
1	'Republicans attempted to pass a resolution that would have enabled Congress to force workers to accept a deal, which was fortunately blocked by (who else) Senator Bernie Sanders.' 'No government ever surrenders power, even its emergency powers—not really.' 'No citizen in a democratic society should want executives from $10trn financial institutions to play a larger role than they already do in defining and implementing social values.'

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_random_sample_svm_head")
# Run inference
preds = model("What could possibly go wrong?")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	3	23.4159	68

Label	Training Sample Count
0	136
1	78

Training Hyperparameters

• batch_size: (8, 8)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• 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
0.0003	1	0.3597	-
0.0161	50	0.2693	-
0.0323	100	0.2501	-
0.0484	150	0.2691	-
0.0645	200	0.063	-
0.0806	250	0.0179	-
0.0968	300	0.0044	-
0.1129	350	0.0003	-
0.1290	400	0.0005	-
0.1452	450	0.0002	-
0.1613	500	0.0003	-
0.1774	550	0.0001	-
0.1935	600	0.0001	-
0.2097	650	0.0001	-
0.2258	700	0.0001	-
0.2419	750	0.0001	-
0.2581	800	0.0	-
0.2742	850	0.0001	-
0.2903	900	0.0002	-
0.3065	950	0.0	-
0.3226	1000	0.0	-
0.3387	1050	0.0002	-
0.3548	1100	0.0	-
0.3710	1150	0.0001	-
0.3871	1200	0.0001	-
0.4032	1250	0.0	-
0.4194	1300	0.0	-
0.4355	1350	0.0	-
0.4516	1400	0.0001	-
0.4677	1450	0.0	-
0.4839	1500	0.0	-
0.5	1550	0.0001	-
0.5161	1600	0.0001	-
0.5323	1650	0.0	-
0.5484	1700	0.0	-
0.5645	1750	0.0	-
0.5806	1800	0.0	-
0.5968	1850	0.0	-
0.6129	1900	0.0	-
0.6290	1950	0.0001	-
0.6452	2000	0.0	-
0.6613	2050	0.0	-
0.6774	2100	0.0	-
0.6935	2150	0.0001	-
0.7097	2200	0.0	-
0.7258	2250	0.0	-
0.7419	2300	0.0001	-
0.7581	2350	0.0001	-
0.7742	2400	0.0001	-
0.7903	2450	0.0	-
0.8065	2500	0.0	-
0.8226	2550	0.0	-
0.8387	2600	0.0	-
0.8548	2650	0.0001	-
0.8710	2700	0.0001	-
0.8871	2750	0.0	-
0.9032	2800	0.0	-
0.9194	2850	0.0	-
0.9355	2900	0.0001	-
0.9516	2950	0.0	-
0.9677	3000	0.0001	-
0.9839	3050	0.0	-
1.0	3100	0.0	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.0.3
• Sentence Transformers: 3.0.1
• Transformers: 4.39.0
• PyTorch: 2.3.0+cu121
• Datasets: 2.20.0
• 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}
}