yizhongw / Tk-Instruct

Tk-Instruct is a Transformer model that is tuned to solve many NLP tasks by following instructions.
https://arxiv.org/abs/2204.07705
MIT License
177 stars 27 forks source link

Predicting a topic that doesnot exist in the list #18

Open tayfuntuna opened 1 year ago

tayfuntuna commented 1 year ago

I was using 3b model and i was framing the model as following:

Definition: Find the topic of the given text among this topic list: topic1,topic2

Postive example 1 Input Output: topic1 Explanation: .....because of that result is topic1 ....

When i ran the model, the model will output a topic that is not in the list in the definition or in the outputs of any example inputs.

What can we do to avoid this?