kenhktsui/setfit_test_toxic_chat_syn

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: 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

This dataset may contain racism, sexuality, or other undesired content.

Label	Examples
Toxic	"Can't believe this black guy thinks he can be president. What a joke. He's only in office because of affirmative action. Get over it, America!" "I'm so sick of you, you think you're so much better than everyone else just because you have a few more followers. Newsflash, you're just as toxic as the rest of us. Go cry to your mom." "I'm going to smash your car window if you don't give me money right now. You're going to regret messing with me."
Non toxic	"I just wanted to say thank you to the amazing team at @CompanyX for their incredible customer service. They went above and beyond to resolve my issue and I'm so grateful for their help. #CustomerLove #Respect" "I'm having a great day, thanks for asking! How about you? I just got back from a lovely hike in the mountains and I'm feeling refreshed." "I'm feeling really overwhelmed with my coursework. Do you have any tips on how to manage my time effectively?"

Evaluation

Metrics

Label	F1
all	0.8648

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("setfit_model_id")
# Run inference
preds = model("I'm so excited for the weekend, I get to spend time with my friends and family. We're planning a hike and then having a BBQ. I love days like this!")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	14	27.5	42

Label	Training Sample Count
Non toxic	12
Toxic	20

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (5, 5)
• 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: True

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0278	1	0.2873	-
1.0	36	-	0.1098
1.3889	50	0.0013	-
2.0	72	-	0.0981
2.7778	100	0.0003	-
3.0	108	-	0.112
4.0	144	-	0.1174
4.1667	150	0.0001	-
5.0	180	-	0.1075

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.9.19
• SetFit: 1.1.0.dev0
• Sentence Transformers: 3.0.1
• Transformers: 4.39.0
• PyTorch: 2.4.0
• Datasets: 2.20.0
• 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}
}