DistilT5 for question-generation

This is distilled version of t5-small-qa-qg-hl model trained for question answering and answer aware question generation tasks.

The model is distilled using the No Teacher Distillation method proposed by Huggingface, here.

We just copy alternating layers from t5-small-qa-qg-hl and finetune more on the same data. Following table lists other distilled models and their metrics.

You can play with the model using the inference API. Here's how you can use it

For QG

generate question: <hl> 42 <hl> is the answer to life, the universe and everything.

For QA

question: What is 42 context: 42 is the answer to life, the universe and everything.

For more deatils see this repo.

Model in action 🚀

You'll need to clone the repo.

from pipelines import pipeline
nlp = pipeline("multitask-qa-qg", model="valhalla/distilt5-qa-qg-hl-6-4")

# to generate questions simply pass the text
nlp("42 is the answer to life, the universe and everything.")
=> [{'answer': '42', 'question': 'What is the answer to life?'}]

# for qa pass a dict with "question" and "context"
nlp({
    "question": "What is 42 ?",
    "context": "42 is the answer to life, the universe and everything."
})
=> 'the answer to life, the universe and everything'