SetFit with thenlper/gte-base

This is a SetFit model that can be used for Text Classification. This SetFit model uses thenlper/gte-base 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: thenlper/gte-base
Classification head: a LogisticRegression 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
True	'Tech Giant A has acquired Startup B in a groundbreaking deal valued at $500 million, aiming to enhance its product offerings.' 'Industry leaders D and E announced their merger today, combining resources to capture more market share in the competitive landscape.' 'International Firm I and Domestic Firm J have finalized an acquisition deal that is expected to reshape the industry.'
False	'In an unexpected move, Company C has invested heavily in a new technology initiative, signaling its commitment to innovation.' 'A recent survey indicates that Company F is planning to introduce new features next quarter, boosting user engagement.' 'Company G has entered a partnership with Organization H, focusing on joint product development and marketing strategies.'

Evaluation

Metrics

Label	Accuracy
all	0.93

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("amplyfi/merger-and-acquisition")
# Run inference
preds = model("The government has announced new regulations on corporate mergers and acquisitions, affecting multiple industries.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	9	14.4496	25

Label	Training Sample Count
False	243
True	213

Training Hyperparameters

batch_size: (16, 16)
num_epochs: (3, 3)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 5
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
l2_weight: 0.01
seed: 42
eval_max_steps: -1
load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0035	1	0.1828	-
0.1754	50	0.2934	-
0.3509	100	0.0797	-
0.5263	150	0.0108	-
0.7018	200	0.0013	-
0.8772	250	0.0007	-
1.0526	300	0.0003	-
1.2281	350	0.0002	-
1.4035	400	0.0002	-
1.5789	450	0.0002	-
1.7544	500	0.0002	-
1.9298	550	0.0002	-
2.1053	600	0.0001	-
2.2807	650	0.0001	-
2.4561	700	0.0001	-
2.6316	750	0.0001	-
2.8070	800	0.0001	-
2.9825	850	0.0001	-

Framework Versions

Python: 3.10.12
SetFit: 1.1.0
Sentence Transformers: 3.3.1
Transformers: 4.42.2
PyTorch: 2.5.1+cu124
Datasets: 3.1.0
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}
}

amplyfi
/

merger-and-acquisition