Smit proposes the following approach:
- Prioritize the lectures that need to be converted to JAX and are doable without external library support.
- Start tackling them one by one in a weekly sprint.
- Once we have some requirement like
brent_max
-- open an issue to track this and move on to the next lecture. - We can do this for let's say 10 lectures, and, then come back to those issues and look for fixing or implementing them.
- Create a new repository for these JAX lectures until we have substantial lectures so that it will not affect current lectures -- considering build issues due to new releases of JAX.
Comments from JS
- JS to-do: prioritize lectures
- Suggest a benchmark example that we can build off
- We will probably think differently about how to approximate and optimize