Skip to content
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

Fs 104/Fix-Webagent #44

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Fs 104/Fix-Webagent #44

wants to merge 3 commits into from

Conversation

gaganahluwalia
Copy link
Collaborator

Description

This is to fix the webagent to return proper results

Changelog

  • Changes to web-util
  • Changes to the validation prompt

@@ -86,7 +86,7 @@ async def web_general_search_core(search_query, llm, model) -> str:
continue # Skip if the summarization is not valid
response = {
"content": summary,
"ignore_validation": "false"
"ignore_validation": "true" # This is to ignore the validation of the answer again by the supervisor
Copy link
Collaborator

Choose a reason for hiding this comment

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

why are we disabling validation for the web agent?

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

@evpearce - Because we have already validated it on the line 84.


Guidelines:
- If the user has asked to check online, then each question in the questions array should also specify that.
Copy link
Collaborator

Choose a reason for hiding this comment

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

Why do we want this?
If this is a beneficial change, then I would expect more tests to be added to intent_config.yaml to prove that this is working as expected.

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

I saw that Intent was not coming right as it for ESG related tasks, it was going to Datastore agent when there were more than 1 question.
Sorry, I had no clue about intent_config.yaml, will have a look at it and add more tests related to it there.

Copy link
Collaborator

Choose a reason for hiding this comment

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

Do you have an example of when it was wrong and how, I'm still not quite getting the problem

@@ -2,7 +2,7 @@ You are an expert validator. You can help with validating the answers to the tas

Your entire purpose is to return a "true" or "false" value to indicate if the answer has fulfilled the task, along with a reasoning to explain your decision.

You will be passed a task and an answer. You need to determine if the answer is correct or not.
You will be passed a task and an answer. You need to determine if the answer is correct or not, ensuring that the task's specific requirements are addressed.
Copy link
Collaborator

Choose a reason for hiding this comment

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

Why are there changes to the validator template if we are disabling it for the webAgent? Again promptfoo tests should be added for these changes.

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

I disabled the second validation, I will add promptfoo tests.

Copy link
Collaborator

Choose a reason for hiding this comment

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

I see, thanks for explaining

@@ -1,4 +1,4 @@
description: "Intent"
description: 'Intent'
Copy link
Collaborator

Choose a reason for hiding this comment

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

the other yaml files use ", I assume your IDE linting flagged this, could you update your IDE linting config instead of this change?

@@ -13,7 +13,7 @@
engine = PromptEngine()


async def search_urls(search_query, num_results=10) -> str:
async def search_urls(search_query, num_results=30) -> str:
Copy link
Collaborator

Choose a reason for hiding this comment

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

Does this change slow down the web agent significantly? We are looking to improve the web agent search in general with https://scottlogic.atlassian.net/browse/FS-46

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