thunlp/OpenPrompt

Health v2 · maintenance only

Adoption not part of the score

4890 stars · 476 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

OpenPrompt is a PyTorch-based open-source framework for prompt-learning, providing a standard, flexible pipeline of templates and verbalizers to adapt pre-trained language models to downstream NLP tasks. It loads PLMs directly from Hugging Face transformers and supports manual, soft, prefix-tuning, and generation-style prompting.

Use cases

• apply prompt-based tuning to pre-trained language models
• run few-shot text classification with templates and verbalizers
• adapt BERT or T5 to downstream NLP tasks via prompting
• experiment with manual and soft prompt templates in PyTorch
• do supervised instruction tuning for a chat model
• benchmark prompt-learning methods on SuperGLUE

When to choose

• you want a research-grade, extensible framework for prompt-learning pipelines
• your models come from Hugging Face transformers and you need template/verbalizer abstractions
• you need few-shot classification or generation paradigms over PLMs

When to avoid

• you need modern LLM fine-tuning with LoRA/PEFT or chat-style inference rather than classic prompt-learning
• you want a library with frequent updates and current model support
• you just need prompt engineering for an LLM API without training

Facets

library · maturity maintenance

machine-learning nlp prompt-engineering llm-training machine-learning deep-learning large-language-models python cross-platform prompt-learning pytorch pre-trained-language-models prompt-tuning huggingface-transformers few-shot-learning natural-language-processing

Member repositories

For agents

markdown · JSON · MCP: product_card(name="thunlp/OpenPrompt")