There is some core modelling work that needs proper attention.
We need folks who are knowledgeable/experienced/motivated to spend an hour or two thinking about a few ways to model this, and what an MVP technical implementation looks like.
I suspect it may be #12 and/or #63, but there may be a more direct and simpler way.
Ultimately we would like to publish a paper on the results.
Useful starting points:
In particular, we want to extend our models to include:
- screen shots (vision)
- structured window data
- structured action data
- natural language task descriptions
- API documentation
Related: #252
There is some core modelling work that needs proper attention.
We need folks who are knowledgeable/experienced/motivated to spend an hour or two thinking about a few ways to model this, and what an MVP technical implementation looks like.
I suspect it may be #12 and/or #63, but there may be a more direct and simpler way.
Ultimately we would like to publish a paper on the results.
Useful starting points:
In particular, we want to extend our models to include:
Related: #252