Skip to content

feat: add evolving history and status to runner #1084

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 22 commits into
base: main
Choose a base branch
from

Conversation

RolandMinrui
Copy link
Collaborator

@RolandMinrui RolandMinrui commented Jul 17, 2025

This pr is intended to:

  • Make the feedback of runner to multi-turn conversation.
  • Add a "role" or "status" description for each of agents. This should includes what does the pipeline do, what's your role in the pipeline, what is the percentage of runtime used and what is the loops used, etc.
  • Add change summary to runner.
  • Add previous loops information to runner.

@RolandMinrui RolandMinrui marked this pull request as draft July 17, 2025 06:18
@@ -79,10 +79,17 @@ def multistep_evolve(
# TODO: Putting the evolving trace in here doesn't actually work
queried_knowledge = self.rag.query(evo, self.evolving_trace)

# 2.5 Brief evolving history
evolving_history = (
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we defer therender it to last step

@@ -92,7 +99,7 @@ def multistep_evolve(
# 5. Evaluation
if self.with_feedback:
es.feedback = (
eva if isinstance(eva, Feedback) else eva.evaluate(evo, queried_knowledge=queried_knowledge)
eva if isinstance(eva, Feedback) else eva.evaluate(evo, evolving_history=evolving_history)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why do we have to remove queried_knowledge=queried_knowledge?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's not used in current runner

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

image

But it is used in other evaluators...

@@ -51,6 +51,13 @@ class EvoStep:
evolvable_subjects: EvolvableSubjects
queried_knowledge: QueriedKnowledge | None = None
feedback: Feedback | None = None
code_change_summary: str | None = None # TODO: minrui
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this place is a good place to add summary,
Let's have the discussion later.

@@ -48,6 +64,7 @@ def evaluate(
implementation: FBWorkspace,
gt_implementation: FBWorkspace,
queried_knowledge: QueriedKnowledge = None,
evolving_history: tuple = None,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

eolving_trace

## Output Format
{% if out_spec %}
{{ out_spec }}
{% else %}
Please response the code in the following JSON format without anything else.
{
"code": "The Python code as a string."
"code": "The refined Python code as a string."
"code_change_summary": "The structured summary to briefly summarize the changes made to the original code base in two to three sentences."
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

remember change

@RolandMinrui RolandMinrui requested a review from peteryang1 July 23, 2025 06:58
@RolandMinrui RolandMinrui marked this pull request as ready for review July 23, 2025 12:49
@RolandMinrui RolandMinrui force-pushed the minrui/add_runner_history branch from fd27cb7 to 1064b68 Compare July 25, 2025 03:13
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants