Improving task classification by using cache and GPT-J along with chatGPT
ahmed3520 opened this issue · comments
This feature proposal aims to improve the accuracy of task classification in our project by leveraging the capabilities of GPT-J and chatGPT, along with the use of cache. By using GPT-J, a small and open-source model, we can quickly classify tasks with high accuracy. In cases where the accuracy is below a certain threshold, we can send the request to chatGPT for further processing. Additionally, by using cache, we can store and retrieve previous classification results to speed up the process. This approach has the potential to significantly improve the performance and accuracy of our task classification system.
I used GPT-J but it requires a lot of RAM that I cannot access for free in Colab. So I decided to go with bloob-3B and did it. Now I will work on measuring the accuracy of bloob-3B reply