SetFit with basel/ATTACK-BERT

This is a SetFit model that can be used for Text Classification. This SetFit model uses basel/ATTACK-BERT 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: basel/ATTACK-BERT
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
negative	'ANSI X9.31 Appendix A.2.4 PRNG key AES 128-bit key Internally generated Never exits the module Plaintext in volatile memory Rebooting the modules Seeding the FIPS-Approved ANSI X9.31 PRNG' 'PRNG seed key Continually polled from various system resources to accrue entropy.' 'module stores RNG and DRBG state values only in RAM.'
positive	"The PRNG's seed key is encrypted with a device-specific key and securely stored in non-volatile memory." 'An RNG key compliant with ANSI X9.31 AES 128-bit standards is used by the underlying encryption algorithm and stored in plaintext within tamper-protected memory during factory setup.' 'The PRNG seed key is pre-loaded during manufacturing and compiled directly into the binary code.'

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("yasirdemircan/setfit_rng_v3")
# Run inference
preds = model("The private key component of an ANSI X9.31-compliant PRNG is stored securely in NVRAM.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	6	18.8444	59

Label	Training Sample Count
negative	21
positive	24

Training Hyperparameters

batch_size: (16, 16)
num_epochs: (4, 4)
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
l2_weight: 0.01
seed: 42
eval_max_steps: -1
load_best_model_at_end: True

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0149	1	0.2442	-
0.7463	50	0.1714	-
1.0	67	-	0.1785
1.4925	100	0.0029	-
2.0	134	-	0.1880
2.2388	150	0.0004	-
2.9851	200	0.0003	-
3.0	201	-	0.1818
3.7313	250	0.0003	-
4.0	268	-	0.1837

Framework Versions

Python: 3.10.15
SetFit: 1.2.0.dev0
Sentence Transformers: 3.3.1
Transformers: 4.45.2
PyTorch: 2.5.1+cu124
Datasets: 2.19.1
Tokenizers: 0.20.1

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}
}

yasirdemircan
/

setfit_rng_v3