richie-ghost/setfit-mobile-bert-phatic

SetFit with csarron/mobilebert-uncased-squad-v2

This is a SetFit model that can be used for Text Classification. This SetFit model uses csarron/mobilebert-uncased-squad-v2 as the Sentence Transformer embedding model. A SetFitHead 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: csarron/mobilebert-uncased-squad-v2
• Classification head: a SetFitHead 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

Label	Examples
1	"How's the family?" 'Thanks a million.' 'I appreciate your kindness.'
0	'What is the next step in the process?' 'Please complete the review by the end of the week.' 'I feel disconnected from reality.'

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("richie-ghost/setfit-mobile-bert-phatic")
# Run inference
preds = model("Have a good day!")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	1	8.2394	184

Label	Training Sample Count
0	143
1	116

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
• l2_weight: 0.01
• seed: 42
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0009	1	0.3528	-
1.0	1068	0.0252	0.0729
2.0	2136	0.0001	0.0544
0.0015	1	0.0	-
0.0772	50	0.001	-
0.1543	100	0.0	-
0.2315	150	0.0	-
0.3086	200	0.0	-
0.3858	250	0.0015	-
0.4630	300	0.001	-
0.5401	350	0.0	-
0.6173	400	0.0	-
0.6944	450	0.0	-
0.7716	500	0.0	-
0.8488	550	0.0	-
0.9259	600	0.0	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.1.0
• Sentence Transformers: 3.2.1
• Transformers: 4.39.0
• PyTorch: 2.0.1+cu117
• Datasets: 3.1.0
• 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}
}