timoschick/pet

• highabout#1

  Clarify the project's full name and domain in the 'About' description

  Why:

  CURRENT

  This repository contains the code for "Exploiting Cloze Questions for Few-Shot Text Classification and Natural Language Inference"

  COPY-PASTE FIX

  Copy

  Code for Pattern-Exploiting Training (PET), a semi-supervised method for few-shot text classification and natural language inference.

  OpenDoneDismiss
• highreadme#2

  Emphasize semi-supervised learning and unlabeled data in the README's introduction

  Why:

  CURRENT

  This repository contains the code for Exploiting Cloze Questions for Few-Shot Text Classification and Natural Language Inference and It's Not Just Size That Matters: Small Language Models Are Also Few-Shot Learners. The papers introduce pattern-exploiting training (PET), a semi-supervised training procedure that reformulates input examples as cloze-style phrases. In low-resource settings, PET and iPET significantly outperform regular supervised training, various semi-supervised baselines and even GPT-3 despite requiring 99.9% less parameters. The iterative variant of PET (iPET) trains multiple generations of models and can even be used without any training data.

  COPY-PASTE FIX

  Copy

  This repository contains the code for Exploiting Cloze Questions for Few-Shot Text Classification and Natural Language Inference and It's Not Just Size That Matters: Small Language Models Are Also Few-Shot Learners. The papers introduce Pattern-Exploiting Training (PET), a powerful semi-supervised training procedure that reformulates input examples as cloze-style phrases. PET and its iterative variant (iPET) are particularly effective for leveraging unlabeled data in low-resource settings, significantly outperforming regular supervised training and various semi-supervised baselines for tasks like text classification and natural language inference, even without any training data.

  OpenDoneDismiss
• mediumtopics#3

  Add more specific topics to improve discoverability for core methodologies

  Why:

  CURRENT

  machine-learning, nlp, python

  COPY-PASTE FIX

  Copy

  machine-learning, nlp, python, few-shot-learning, semi-supervised-learning, text-classification, natural-language-inference

  OpenDoneDismiss

Category GEO backends resolved for this scan: google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• CATEGORY QUERY

  What are effective methods for few-shot text classification in low-resource settings?

  you: #2
  AI recommended (in order):

  1. SetFit
  2. PET ← you
  3. P-tuning v2
  4. AdapterHub
  5. BERT
  6. RoBERTa
  7. Sentence-BERT
  8. XLM-R
  9. Logistic Regression
  10. SVM
  11. MLP
  12. GPT-3.5
  13. GPT-4
  14. SimCSE
  Show full AI answer
• CATEGORY QUERY

  How to leverage unlabeled data for natural language inference without much supervision?

  you: not recommended
  AI recommended (in order):

  1. SetFit
  2. Snorkel
  3. Hugging Face Transformers
  4. Sentence Transformers
  5. Hugging Face Transformers
  6. modAL

  AI recommended 6 alternatives but never named timoschick/pet. This is the gap to close.

  Show full AI answer

• Metadata completeness

  pass

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of timoschick/pet?

  pass
  ✓AI named timoschick/pet explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• If a team adopts timoschick/pet in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI named timoschick/pet explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• In one sentence, what problem does the repo timoschick/pet solve, and who is the primary audience?

  pass
  ✓AI named timoschick/pet explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?