SetFit with Omar-Nasr/setfitmodel

This is a SetFit model that can be used for Text Classification. This SetFit model uses Omar-Nasr/setfitmodel 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:

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
• Sentence Transformer body: Omar-Nasr/setfitmodel
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 256 tokens
• Number of Classes: 4 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
1.0	' Go out for a walk once a day additionally and slowly start increasing the time you spend outside Go out for a walk once a day additionally and slowly start increasing the time you spend outside Start doing sport, either outdoors or at a gym If you can, try to take your dog to a dog park or something like that' ' Try challenging yourself more, take a walk in the park, small things like that make you better Try challenging yourself more, take a walk in the park, small things like that make you better ' " Now I'm not saying to go to a party on the spot, just go out, shop, take a walk in the park, that kind of thing Now I'm not saying to go to a party on the spot, just go out, shop, take a walk in the park, that kind of thing"
2.0	' I’m an equestrian, so I ride horses and manage for a pretty famous trainer I can hold a non work related conversation with a stranger while I’m working but if I met that same person outside of the work day I’d have a panic attack and not be able to say a word' ' On long walks to errands, and whilst power walking for exercise' ' She said no, but we have a tasty forest fruit mix cake I felt high as a kite walking home'
0.0	' Good to know that some people are in the same camp' " I'm sure if the worlds ever did clash that your friends would understand (few people actually enjoy being at work) and, worst case scenario, your coworkers would be surprised at your outgoing nature while around friends" ' If anything you should be thinking about wearing sun screen so you retain your good skin as it becomes your ally as you age outside'
3.0	" While I ended up making progress, it wasn't as fast as I had hoped and I still had a lot of trouble doing some things (such as jogging in public)" ' One, frack you guys who say “just get over it”, you’ve probably never dealt with anxiety, it’s like you are carrying the weight of everyone’s judgements and eyes on you with every possibility of any and every event running through your head all the time I am trying, I force myself outside and to interact but it’s terrifying and people just don’t seem to get that' " I want to go swimming, anxiety and low self esteem make it really hard I want to go swimming, anxiety and low self esteem make it really hard I'm at least planning to go for a swim at a nearby lake but there is one problem I have: I'm not really confident with my body"

Evaluation

Metrics

Label	Accuracy
all	0.5867

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("Omar-Nasr/setfitmodel")
# Run inference
preds = model(" Want to join soccer club but so scared")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	4	51.2656	1083

Label	Training Sample Count
0.0	16
1.0	16
2.0	16
3.0	16

Training Hyperparameters

• batch_size: (8, 8)
• num_epochs: (1, 1)
• 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
• seed: 42
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0026	1	0.0	-
0.1302	50	0.0001	-
0.2604	100	0.0	-
0.3906	150	0.0	-
0.5208	200	0.0	-
0.6510	250	0.0	-
0.7812	300	0.0	-
0.9115	350	0.0	-

Framework Versions

• Python: 3.10.13
• SetFit: 1.0.3
• Sentence Transformers: 2.7.0
• Transformers: 4.39.3
• PyTorch: 2.1.2
• Datasets: 2.18.0
• 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}
}