bhaskars113/guinness-segments-model

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:

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: sentence-transformers/paraphrase-mpnet-base-v2
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 512 tokens
• Number of Classes: 3 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	'I don’t drink much but I like wine tasting. We usually buy local wine to take to dinners and such as we rarely drink wine. (NGL, I am a beer drinker so I’m probably just pleb.)' 'Never do the bottom right 2 again, give you major banker/tech bro adult frat boy vibes. I can see you chugging a beer and talking about bitcoin with those looks. Upper left makes you look younger and great.' 'NGL I like pepsi much more than coke. I dunno why.'
2	'?? angolbryggeri - Hazy Crazy\n\n✴️ IPA\n\n?? Sweden ????\n\n??Abv 6.5%\n\n⭐️ 3.60 / 5.0 ~ avg 3.67\n\n?? systembolaget\n\n#beer #bier #birra #öl #cerveza #øl #craftbeer #ipa #dipa #tipa #sour #gose #berlinerweisse #paleale #pilsner #lager #stout #beeroftheday #beerphotografy #hantverksöl #untappd #beergeek #beerlover #ilovebeer #cheers #beerstagram #instabeer #beerporn #ängöl #sweden' "I'm a feast kind of guy Bring out the roast pig and Flagons of ale" '“Just grab me a beer” legend'
0	"My boys (Aged 20 and 26) have moved out so I can't say what they do in their own homes but when they lived with us they were supper straight laced and had no desire to explore Alcohol or Drugs. They were into Gaming or Sports not Partying. Weed is Legal here and as far as I know they are not into that either. They definitely don't smoke, maybe they do Gummies but that would be about it." "Like you said cost is a big one. Plus I just wonder if younger generations might not be into it as much. I can't remember the beer company, but one is talking about making a non alcoholic drink, since the younger generation aren't drinking beer as much. " 'She just graduated and I know they drink occasionally, but it’s all Mike’s Lemonade and White Claw city. Very tame stuff. Her friend group also experimented with that fake pot stuff, I forget the name. I told her I wasn’t okay with that and I’d buy her actual pot (rec is legal in my state) if she was determined to try it, but they apparently all lost interest.'

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("bhaskars113/guinness-segments-model")
# Run inference
preds = model("Can Earned the Brewery Pioneer (Level 6) badge! Earned the I Believe in IPA! (Level 5) badge!")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	6	45.7143	135

Label	Training Sample Count
0	14
1	14
2	14

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (1, 1)
• 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
0.0095	1	0.2908	-
0.4762	50	0.0394	-
0.9524	100	0.0021	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.0.1
• Sentence Transformers: 2.2.2
• Transformers: 4.35.2
• PyTorch: 2.1.0+cu121
• Datasets: 2.16.1
• Tokenizers: 0.15.0

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}
}