Skip to content

feat: add LMEB memory-type grouping leaderborad#4914

Open
cosylsine wants to merge 1 commit into
embeddings-benchmark:mainfrom
cosylsine:feat/lmeb-leaderboard-grouping
Open

feat: add LMEB memory-type grouping leaderborad#4914
cosylsine wants to merge 1 commit into
embeddings-benchmark:mainfrom
cosylsine:feat/lmeb-leaderboard-grouping

Conversation

@cosylsine

Copy link
Copy Markdown

Summary

This PR adds memory-type benchmark groupings for LMEB:

  • LMEB Episodic Memory
  • LMEB Dialogue Memory
  • LMEB Semantic Memory
  • LMEB Procedural Memory

The existing aggregate LMEB benchmark is kept unchanged.

Closes #4898.

Changes

  • Added four LMEB memory-type Benchmark definitions.
  • Added leaderboard menu entries for the new LMEB memory groups.
  • Reused existing LMEB tasks.

Task mapping

Memory Type Tasks
Episodic ...
Dialogue ...
Semantic ...
Procedural ...

Tests

  • make lint
  • make typecheck
  • make test

Copilot AI review requested due to automatic review settings July 13, 2026 04:45

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

Pull request overview

This PR introduces four LMEB sub-benchmarks grouped by memory type (Episodic, Dialogue, Semantic, Procedural) while keeping the existing aggregate LMEB benchmark intact, and wires these groupings into the leaderboard navigation.

Changes:

  • Refactors LMEB task lists into memory-type groupings and defines four new Benchmark instances.
  • Updates benchmark exports to include the new LMEB group benchmarks.
  • Adds leaderboard menu entries to surface the LMEB groups under Retrieval/Domain.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File Description
mteb/benchmarks/benchmarks/benchmarks.py Adds LMEB memory-type task groupings and corresponding new benchmark definitions.
mteb/benchmarks/benchmarks/__init__.py Exposes the new LMEB benchmark objects for registry/import use.
mteb/benchmarks/_leaderboard_menu.py Adds menu entries to display LMEB and its memory-type sub-benchmarks.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

)

LMEB_EPISODIC = Benchmark(
name="LMEB-Episodic",
)

LMEB_DIALOGUE = Benchmark(
name="LMEB-Dialogue",
)

LMEB_SEMANTIC = Benchmark(
name="LMEB-Semantic",
)

LMEB_PROCEDURAL = Benchmark(
name="LMEB-Procedural",

@Samoed Samoed left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

I thought you'll update to have one benchmark view aggregations per task, but this will also work I think

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.

Add memory-type grouping for LMEB tasks: episodic, dialogue, semantic, and procedural memory

3 participants