You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
What problem are we solving?
There are many agent workflows where one generation is simply not enough.
For example:
Generate → Review → Improve → Review again
Today, developers have to build this loop themselves. We could provide a reusable pattern for it in MAF.
Proposed idea
Add an IterativeBuilder that keeps running the workflow until the output passes a quality check or reaches the maximum number of iterations.
Example
workflow = (
IterativeBuilder()
.generate(generator)
.evaluate(evaluator)
.improve(improver)
.max_iterations(3)
.build()
)
The main idea is simple:
Generate something, check its quality, improve it if needed, and stop when it is good enough.
This could be useful for:
Draft → Critique → Rewrite
Code → Test → Fix
Generate → Validate → Refine
RAG Answer → Evaluate → Improve
It gives developers a common way to implement these iterative workflows instead of creating the loop manually each time.
All reactions