-
Notifications
You must be signed in to change notification settings - Fork 8
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Added first pass at getting started with qna for wikipedia knowledge submissions Signed-off-by: JJ Asghar <[email protected]>
- Loading branch information
Showing
2 changed files
with
58 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,57 @@ | ||
In this tutorial we will walk you through building out a new `qna.yaml` for adding new or updated knowledge to the `granite-7b-lab` model. Let's get started! | ||
|
||
```bash | ||
mkdir instructlab | ||
git clone [email protected]:erictherobot/wikipedia-markdown-generator.git | ||
``` | ||
The first thing we need to do is create a new directory to have a clean place to work and pull down some software. Most of the time, the easiest thing to update in the model is the Wikipedia entry, so luckily, `erictherobot` has written a helpful tool to pull down markdown versions of the articles for us. | ||
|
||
```bash | ||
git clone [email protected]:<USERNAME>/instructlab-knowledge-docs.git | ||
``` | ||
|
||
After this, clone down your instructlab knowledge docs repository. It can be named whatever you'd like, but if you use our https://ui.instructlab.ai, you'll notice you already have `instructlab-knowledge-docs`. | ||
|
||
```bash | ||
cd wikipedia-markdown-generator | ||
python3.11 -m venv venv-md-gen | ||
source venv-md-gen/bin/activate | ||
pip install -r requirements | ||
python3 wiki-to-md.py Texas_Longhorns_football | ||
``` | ||
Next, we need to build a Python virtual environment and install the dependencies to get it to work. These commands cd into the directory, create the virtual environment with python3.11 (you may need to change the version of Python on your machine), activate the virtual environment, and then do the pip install the dependencies. | ||
You'll notice the `Texas_Longhorns_football` there, a Wikipedia article I wanted to pull down and create the `qna.yaml` against. You should choose whatever new knowledge you want to do here. | ||
|
||
```bash | ||
cp md_output/Texas_Longhorns_football.md ../../instructlab-knowledge-docs/ | ||
cd ../../instructlab-knowledge-docs | ||
git add . | ||
git commit -m "added markdown doc" | ||
git push origin main | ||
cd .. | ||
``` | ||
|
||
Next, we go ahead and copy the markdown into the knowledge repository, and commit it to our repository and push it up to GitHub. | ||
|
||
``` | ||
git clone [email protected]/instructlab/taxonomy | ||
cd taxonomy | ||
``` | ||
|
||
Next we pull down the upstream public taxonomy directory, and `cd` into that directory. | ||
|
||
```bash | ||
mkdir -p arts/sports/american_football/college/university_of_texas/ | ||
``` | ||
This next step is a "best effort" for you. As the taxonomy grows, there will be some obvious choices, but if you select a tree that hasn't been flushed out yet, you'll have to do your best to think about where you'd find the `qna.yaml`. In this case, the Dewey Decimal System says sports should be under arts; this is American Football, college level with the University of Texas. Also, notice the underscores for the spaces; this is important. | ||
|
||
```bash | ||
wget https://raw.githubusercontent.com/instructlab/taxonomy/main/docs/template_qna.yaml | ||
mv template_qna.yaml sports/american_football/college/university_of_texas/qna.yaml | ||
``` | ||
Finally, you can pull down the `template_qna.yaml` and fill it out for the needed questions and answers. Be sure to put the context at a maximum of about 500 Tokens and questions and answers around 250 Tokens. | ||
|
||
``` | ||
vim sports/american_football/college/university_of_texas/qna.yaml | ||
``` | ||
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters