Would love to see the following improvements for playground as I use it all the time:
- Better support for mobile (or even an app)
- Minimum width of the input area in edit mode. If the output has really wide sentences, the input window is sometimes “squashed” to something really narrow.
- Often when switching between edit mode and Complete mode the engine is not set accordingly, and I get a red warning that I am using the wrong endpoint.
The playground is so useful in itself already, and I suspect the more this advances, the less we will need to pre prompt or pre-train for specific tasks. This will just be a multi functional single interface to the AI. The most we will have to do is do some one shot examples to hint at an output format wished for.