Skip to content
This repository was archived by the owner on Sep 29, 2025. It is now read-only.

(EAI-1231): Generate NL2AS dataset#898

Merged
mongodben merged 38 commits intomainfrom
EAI-1231
Aug 28, 2025
Merged

(EAI-1231): Generate NL2AS dataset#898
mongodben merged 38 commits intomainfrom
EAI-1231

Conversation

@mongodben
Copy link
Copy Markdown
Collaborator

@mongodben mongodben commented Aug 25, 2025

Jira: https://jira.mongodb.org/browse/EAI-1231

Changes

  • Generate Atlas Search Dataset: atlas-search-dataset-gpt-5
  • Prompt refinement on generate Atlas Search dataset pipeline
  • More robust err handling in tool calls
  • Add GPT-5 and Claude 4 models to benchmarking set. Add dataset to benchmark.

Notes

@mongodben mongodben marked this pull request as ready for review August 25, 2025 20:29
…search_claude-opus-4-20250514_temp_0.9_1755190427429.jsonl
Copy link
Copy Markdown
Collaborator

@hschawe hschawe left a comment

Choose a reason for hiding this comment

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

approved, just a couple of small cleanups

async task(input) {
const promptMessages = makeGenerateNaturalLanguageQueryPrompt({
...input,
queryType: "atlas_search", // Ensure we use Atlas Search prompts
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

nit - could make this a named constant since this same string is used twice in this file

},
] as const satisfies DatabaseUser[];

// TODO: update
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

what's to update / should this be removed?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

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

should be removed. thx

Comment on lines -258 to +311
if (
fastestMostFrequentIndex !== null &&
dbExecutions[fastestMostFrequentIndex].data.result !== null
) {
const dbResult = dbExecutions[fastestMostFrequentIndex].data.result;
if (dbResult && Array.isArray(dbResult) && dbResult.length > 0) {
dbExecutions[fastestMostFrequentIndex].data.isReferenceAnswer = true;
if (fastestMostFrequentIndex !== null) {
const execution = dbExecutions[fastestMostFrequentIndex];
if (execution.data.result !== null) {
const dbResult = execution.data.result;
if (dbResult && Array.isArray(dbResult) && dbResult.length > 0) {
execution.data.isReferenceAnswer = true;
}
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

i prefer the old version of this unless there's a reason to make it more nested

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

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

there were some logic changes in here i believe

@mongodben mongodben merged commit c77e480 into main Aug 28, 2025
2 checks passed
@mongodben mongodben deleted the EAI-1231 branch August 28, 2025 20:10
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants