SetFit with sentence-transformers/paraphrase-mpnet-base-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses sentence-transformers/paraphrase-mpnet-base-v2 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: sentence-transformers/paraphrase-mpnet-base-v2
Classification head: a LogisticRegression instance
Maximum Sequence Length: 512 tokens
Number of Classes: 9 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
business and completing tasks	'I am feeling positive today that I am going to complete as much work as I can to ensure that I can go to work tomorrow, barring exhaustion. I am also excited for the upcoming storm. Storms bring a sense of positivity.' 'When I first started my online store selling books. I thought positive I am going to sell tons of books this is going to be easy work and I am going to make thousands. lol I believed in myself a lot more then.'
connecting with others	"One time that I felt a positive emotion was last week. I was able to see my entire extended family on thanksgiving at my grandmother's house. I just felt overjoyed and filled with love. We haven't had everyone together since before COVID, so it felt great to be around fun, family, friends, and food." "I felt a positive emotion recently when I was at a friend's wedding. During the ceremony, I felt strong emotions of happiness, pride, and love. I felt these emotions because it was so powerful seeing my friends of many years getting married, and hearing them express their love to each other."
dreams and goals	'I feel position when I accomplish a goal or make progress on a goal that I have set for myself. For example, I have a daily goal of walking five miles. If I walk around four or more miles, I feel positive about my day. If I walk more than five miles, I feel even more positive about my accomplishments. ' 'One time, I felt an overwhelming sense of joy, contentment, and gratitude when I was accepted into my dream university. This positive emotion arose from the realization of achieving a long-held goal and the validation of my hard work paying off. I felt an immense sense of pride and excitement about the opportunities that lay ahead, and it motivated me to embark on a new chapter in my life with enthusiasm and determination.'
engaging with hobbies and accomplishments	"Well, this may not be what you're looking for, but I've been feeling happy and enthusiastic about building a new desktop computer. I've ordered the parts and every time one of them comes in, I'm that much closer to the goal. The anticipation isn't really an emotion I suppose, but it is a really positive feeling for me." 'I just felt so excited that I managed to make two fingerless gloves on my knitting looms for the first time. They look and feel great and my mom is going to love knowing I was thinking of her. '
overcoming challenges	'I was able to cut my taxes in half. Also, our homeowners insurance was reduced by almost 1k and we are now receiving more coverage. Additionally, I managed to get our mortgage reduced from $2700 to $603.37. Quite proud of my effort(s) and the results. :)' 'I was able to cut my taxes in half. Also, our homeowners insurance was reduced by almost 1k and we are now receiving more coverage. Additionally, I managed to get our mortgage reduced from $2700 to $603.37. Quite proud of my effort(s) and the results. :)'
parenthood, taking care of something	"This morning I was snuggling with my 9-year-old son. For a few minutes I really looked at his face, at how he's getting older, but how much I still love him. I felt grateful that I have him, a lot of love, and at peace." 'I felt a positive emotion at the birth of my daughter. I was almost 50 at the time and after raising two sons, I knew I was entering, very possibly, a new enlightening and respectful period in my life. As time has passed since then, I have found that love has truly entered my life as never expected.'
professional and academic accomplishments	'I felt a positive emotion when I got promoted as a manager of my firm. I worked really hard to attain this goal. My emotions went out of control when I took charge as a manager of my firm.' 'I felt a surge of confidence and competence when I got my first real job. This real job was based on my hard work at school and was a career job that paid well. I felt my life making a turn to the good and that I could finally relax and feel some energy and peace that I could count on to last a long time.'
quality time and vacations	'I felt a positive emotion when I was on vacation in Hawaii. When i sit on the beach and stare out at the ocean I have a sense of calm and I feel postivie about all the world. I feel in awe of the world and the vast ocean. ' 'I felt a positive emotion when I went to visit NYC recently because I love that city. I find the city to be very exciting and motivating so it brings out many positive emotions in me when I am there.'
simple joys	"I last felt a positive emotion this morning. I go outside every morning into my backyard with my cats, and I watched my cat chase birds unsuccessfully for a few minutes, which had me laughing. He's so cute when he does that, it made my morning." "Thankfulness emerges when we recognize that someone or something is a positive in our life. We might feel gratitude for gifts we've received, kindnesses extended to us, or for something as simple as being able to wake up each day."

Evaluation

Metrics

Label	Accuracy
all	0.4773

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("knharris4/harris")
# Run inference
preds = model("I felt happy and content last night. I was with my husband and daughter and we had just had dinner. We were hanging out, watching tv, eating cookies and playing games. It was amazing!")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	39	50.2222	73

Label	Training Sample Count
business and completing tasks	2
connecting with others	2
dreams and goals	2
engaging with hobbies and accomplishments	2
overcoming challenges	2
parenthood, taking care of something	2
professional and academic accomplishments	2
quality time and vacations	2
simple joys	2

Training Hyperparameters

batch_size: (16, 16)
num_epochs: (2, 2)
max_steps: -1
sampling_strategy: oversampling
num_iterations: 15
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
0.0294	1	0.0416	-
1.4706	50	0.038	-

Framework Versions

Python: 3.10.12
SetFit: 1.1.0
Sentence Transformers: 3.2.1
Transformers: 4.44.2
PyTorch: 2.5.0+cu121
Datasets: 3.0.2
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}
}

knharris4
/

harris