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 Sources

Model Labels

Label Examples
product policy
  • 'Do you offer a gift wrapping service for sneakers?'
  • 'What are the consequences if my account is suspended or terminated for any reason?'
  • 'Do you share my personal information with third parties?'
general faq
  • 'Can you explain why Mashru silk is considered more comfortable to wear compared to pure silk sarees?'
  • 'What are some tips for maximizing the antioxidant content when brewing green tea?'
  • 'Can you recommend K-beauty products for hot and humid climates?'
product discoverability
  • 'Are there any sarees with Kadwa Weave technique?'
  • 'cookie boxes with dividers'
  • 'Are there any products for dry skin?'
Out of Scope
  • 'Is this website secure?'
  • 'How do you handle intellectual property disputes?'
  • 'Do you know how to play the piano?'
order tracking
  • 'I want to deliver candle supplies to Jaipur, how many days will it take to deliver?'
  • 'I want to deliver bags to Pune, how many days will it take to deliver?'
  • 'I need to change the delivery address for my recent order, how can I do that?'
product faq
  • 'Does this product help with dark spots?'
  • '3. Is this product currently in stock?'
  • 'Is the product in stock?'

Evaluation

Metrics

Label Accuracy
all 0.8711

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 like to listen to classical music")

Training Details

Training Set Metrics

Training set Min Median Max
Word count 4 10.66 28
Label Training Sample Count
Out of Scope 50
general faq 50
order tracking 50
product discoverability 50
product faq 50
product policy 50

Training Hyperparameters

  • batch_size: (16, 16)
  • num_epochs: (2, 2)
  • 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
0.0002 1 0.2592 -
0.0107 50 0.2424 -
0.0213 100 0.1506 -
0.0320 150 0.222 -
0.0427 200 0.1227 -
0.0533 250 0.1801 -
0.0640 300 0.1111 -
0.0747 350 0.0346 -
0.0853 400 0.0313 -
0.0960 450 0.0048 -
0.1067 500 0.0023 -
0.1173 550 0.0018 -
0.1280 600 0.0133 -
0.1387 650 0.0008 -
0.1493 700 0.0006 -
0.1600 750 0.0005 -
0.1706 800 0.0008 -
0.1813 850 0.0007 -
0.1920 900 0.0006 -
0.2026 950 0.0006 -
0.2133 1000 0.0003 -
0.2240 1050 0.0026 -
0.2346 1100 0.0004 -
0.2453 1150 0.0004 -
0.2560 1200 0.0004 -
0.2666 1250 0.0005 -
0.2773 1300 0.0005 -
0.2880 1350 0.0003 -
0.2986 1400 0.0001 -
0.3093 1450 0.0001 -
0.3200 1500 0.0002 -
0.3306 1550 0.0002 -
0.3413 1600 0.0002 -
0.3520 1650 0.0001 -
0.3626 1700 0.0004 -
0.3733 1750 0.0002 -
0.3840 1800 0.0005 -
0.3946 1850 0.0002 -
0.4053 1900 0.0002 -
0.4160 1950 0.0001 -
0.4266 2000 0.0001 -
0.4373 2050 0.0001 -
0.4480 2100 0.0001 -
0.4586 2150 0.0001 -
0.4693 2200 0.0002 -
0.4799 2250 0.0048 -
0.4906 2300 0.0001 -
0.5013 2350 0.001 -
0.5119 2400 0.0002 -
0.5226 2450 0.0002 -
0.5333 2500 0.0001 -
0.5439 2550 0.0001 -
0.5546 2600 0.0001 -
0.5653 2650 0.0001 -
0.5759 2700 0.0001 -
0.5866 2750 0.0001 -
0.5973 2800 0.0001 -
0.6079 2850 0.0001 -
0.6186 2900 0.0001 -
0.6293 2950 0.0001 -
0.6399 3000 0.0001 -
0.6506 3050 0.0001 -
0.6613 3100 0.0001 -
0.6719 3150 0.0001 -
0.6826 3200 0.0001 -
0.6933 3250 0.0001 -
0.7039 3300 0.0001 -
0.7146 3350 0.0001 -
0.7253 3400 0.0001 -
0.7359 3450 0.0001 -
0.7466 3500 0.0001 -
0.7573 3550 0.0001 -
0.7679 3600 0.0001 -
0.7786 3650 0.0001 -
0.7892 3700 0.0001 -
0.7999 3750 0.0001 -
0.8106 3800 0.0001 -
0.8212 3850 0.0 -
0.8319 3900 0.0001 -
0.8426 3950 0.0001 -
0.8532 4000 0.0001 -
0.8639 4050 0.0001 -
0.8746 4100 0.0001 -
0.8852 4150 0.0 -
0.8959 4200 0.0001 -
0.9066 4250 0.0001 -
0.9172 4300 0.0001 -
0.9279 4350 0.0001 -
0.9386 4400 0.0001 -
0.9492 4450 0.0001 -
0.9599 4500 0.0001 -
0.9706 4550 0.0001 -
0.9812 4600 0.0 -
0.9919 4650 0.0001 -
1.0026 4700 0.0 -
1.0132 4750 0.0001 -
1.0239 4800 0.0001 -
1.0346 4850 0.0001 -
1.0452 4900 0.0001 -
1.0559 4950 0.0001 -
1.0666 5000 0.0 -
1.0772 5050 0.0 -
1.0879 5100 0.0001 -
1.0985 5150 0.0 -
1.1092 5200 0.0 -
1.1199 5250 0.0 -
1.1305 5300 0.0001 -
1.1412 5350 0.0001 -
1.1519 5400 0.0 -
1.1625 5450 0.0001 -
1.1732 5500 0.0001 -
1.1839 5550 0.0002 -
1.1945 5600 0.0 -
1.2052 5650 0.0 -
1.2159 5700 0.0 -
1.2265 5750 0.0 -
1.2372 5800 0.0001 -
1.2479 5850 0.0001 -
1.2585 5900 0.0001 -
1.2692 5950 0.0 -
1.2799 6000 0.0 -
1.2905 6050 0.0 -
1.3012 6100 0.0001 -
1.3119 6150 0.0 -
1.3225 6200 0.0 -
1.3332 6250 0.0 -
1.3439 6300 0.0 -
1.3545 6350 0.0 -
1.3652 6400 0.0 -
1.3759 6450 0.0 -
1.3865 6500 0.0 -
1.3972 6550 0.0 -
1.4078 6600 0.0 -
1.4185 6650 0.0 -
1.4292 6700 0.0 -
1.4398 6750 0.0 -
1.4505 6800 0.0 -
1.4612 6850 0.0 -
1.4718 6900 0.0001 -
1.4825 6950 0.0001 -
1.4932 7000 0.0 -
1.5038 7050 0.0 -
1.5145 7100 0.0001 -
1.5252 7150 0.0001 -
1.5358 7200 0.0001 -
1.5465 7250 0.0001 -
1.5572 7300 0.0 -
1.5678 7350 0.0 -
1.5785 7400 0.0 -
1.5892 7450 0.0001 -
1.5998 7500 0.0 -
1.6105 7550 0.0 -
1.6212 7600 0.0 -
1.6318 7650 0.0 -
1.6425 7700 0.0 -
1.6532 7750 0.0 -
1.6638 7800 0.0 -
1.6745 7850 0.0 -
1.6852 7900 0.0 -
1.6958 7950 0.0 -
1.7065 8000 0.0 -
1.7172 8050 0.0 -
1.7278 8100 0.0 -
1.7385 8150 0.0001 -
1.7491 8200 0.0 -
1.7598 8250 0.0 -
1.7705 8300 0.0 -
1.7811 8350 0.0001 -
1.7918 8400 0.0 -
1.8025 8450 0.0 -
1.8131 8500 0.0 -
1.8238 8550 0.0 -
1.8345 8600 0.0001 -
1.8451 8650 0.0 -
1.8558 8700 0.0 -
1.8665 8750 0.0001 -
1.8771 8800 0.0 -
1.8878 8850 0.0 -
1.8985 8900 0.0 -
1.9091 8950 0.0001 -
1.9198 9000 0.0 -
1.9305 9050 0.0 -
1.9411 9100 0.0 -
1.9518 9150 0.0 -
1.9625 9200 0.0 -
1.9731 9250 0.0 -
1.9838 9300 0.0 -
1.9945 9350 0.0 -

Framework Versions

  • Python: 3.10.16
  • SetFit: 1.0.3
  • Sentence Transformers: 2.7.0
  • Transformers: 4.40.2
  • PyTorch: 2.2.2
  • Datasets: 2.19.1
  • 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}
}
Downloads last month
27
Safetensors
Model size
109M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for Shankhdhar/classifier_woog_base_oos_combined

Finetuned
(285)
this model

Evaluation results