The dataset was created using an LLM(Mistral 7B Instruct), providing him the content(e.g. A plugin’s docs or a page from the documentation) and asking him to generate a QA on that content
Prepared an evaluation pipeline
Ready to evaluate and tune
First results show that as expected the retrieval have to be optimized using a hybrid approach
Working with Notebooks to exploit Colab/Kaggle
Improving Tekton Client Plugin for Jenkins - Maeve Ho
These marked the end of full updates to latest state of jenkins.io in docs.jenkins.io repo . Although this is just part 1 of our Antora work . We will come back to the antora part like migration of pages and division again from July 1.
Also previously part 7 and 8 were in draft mode since some more changes were needed . Now part 8 is ready and part 7 will be completely ready for review by tonight.
I will create some sub-issues inside this main issue for some of the sub pages inside download folder that needs to be migrated as well as decided in our yesterday project meet.