SetFit with intfloat/multilingual-e5-large

This is a SetFit model that can be used for Text Classification. This SetFit model uses intfloat/multilingual-e5-large 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: intfloat/multilingual-e5-large
Classification head: a LogisticRegression instance
Maximum Sequence Length: 512 tokens
Number of Classes: 7 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
6	'What kind of promotions generally lead to higher cannibalization?' 'Which Skus has higher Canninibalization in Natural Juices for 2023?' 'Which two Product can have simultaneous Promotions?'
2	'Which Promotions contributred the most lift Change between 2022 and 2023?' 'Which category x brand has seen major decline in Volume Lift for 2023?' 'What actions were taken to increase the volume lift for MEGAMART in 2023?'
3	'What types of promotions within the FIZZY DRINKS category are best suited for offering high discounts?' 'Which promotion types are better for high discounts in Hydra category for 2022?' 'Which promotion types in are better for low discounts in FIZZY DRINKS category?'
5	'How will increasing the discount by 50 percent on Brand BREEZEFIZZ affect the incremental volume lift?' 'How will the introduction of a 20% discount promotion for Rice Krispies in August affect incremental volume and ROI?' 'If I raise the discount by 20% on Brand BREEZEFIZZ, what will be the incremental roi?'
0	'For which category MULTISAVING type of promotions worked best for WorldMart in 2022?' 'What type of promotions worked best for WorldMart in 2022?' 'Which subcategory have the highest ROI in 2022?'
4	'Suggest a better investment strategy to gain better ROI in 2023 for FIZZY DRINKS' 'Which promotions have scope for higher investment to drive more ROIs in UrbanHub ?' 'What promotions in FIZZY DRINKS have shown declining effectiveneHydra and can be discontinued?'
1	'How do the performance metrics of brands in the FIZZY DRINKS category compare to those in HYDRA and NATURAL JUICES concerning ROI change between 2021 to 2022?' 'Can you identify the specific factors or challenges that contributed to the decline in ROI within ULTRASTORE in 2022 compared to 2021?' 'What are the main reasons for ROI decline in 2022 compared to 2021?'

Evaluation

Metrics

Label	Accuracy
all	1.0

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("vgarg/promo_prescriptive_29_04_2024")
# Run inference
preds = model("Which promotion types are better for low discounts for Zucaritas ?")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	7	14.6667	27

Label	Training Sample Count
0	10
1	10
2	10
3	10
4	10
5	10
6	9

Training Hyperparameters

batch_size: (16, 16)
num_epochs: (3, 3)
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
0.0058	1	0.3528	-
0.2890	50	0.0485	-
0.5780	100	0.0052	-
0.8671	150	0.0014	-
1.1561	200	0.0006	-
1.4451	250	0.0004	-
1.7341	300	0.0005	-
2.0231	350	0.0004	-
2.3121	400	0.0004	-
2.6012	450	0.0005	-
2.8902	500	0.0004	-

Framework Versions

Python: 3.10.12
SetFit: 1.0.3
Sentence Transformers: 2.7.0
Transformers: 4.40.0
PyTorch: 2.2.1+cu121
Datasets: 2.19.0
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}
}

vgarg
/

promo_prescriptive_29_04_2024