GeezerVison

Purpose of the change

Four evaluation scripts crash before doing any work: they unpack three names from init_memmachine_params, which has returned five values since #1376 (evaluation/utils/agent_utils.py:523 returns memory, answer_model, query_agent, agent_model_id, answer_model_id). Each of these sites raises ValueError: too many values to unpack (expected 3):

• evaluation/episodic_memory/locomo_ingest.py:61
• evaluation/episodic_memory/locomo_search.py:177
• evaluation/episodic_memory/locomo_delete.py:52
• evaluation/retrieval_agent/longmemeval_test.py:260 (longmemeval_delete)

Description

Each of the four call sites now unpacks the five values the function returns, binding the unused ones to _. No other change: the call sites under evaluation/retrieval_agent/ and evaluation/retrieval_agent/beam/ already unpack all five.

Fixes/Closes

Fixes #1911

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

• Test Script (please provide)
  • An ast-based check that derives the function's arity from its return statement and compares it with the tuple target at every call site under evaluation/ (18 sites), run in a python:3.12-slim container: before the change it reports the four mismatches above; after, call sites scanned: 18, mismatches: 0.
  • ruff check evaluation/ and ruff format --check on the four changed files with the pinned 0.16.8: clean.
  • A full run of the LoCoMo scripts needs a configured server and model access, and evaluation/ has no test suite and is outside what CI type-checks, so no unit test is added.

Checklist

• I have signed the commit(s) within this pull request
• My code follows the style guidelines of this project (See STYLE_GUIDE.md)
• I have performed a self-review of my own code
• My changes generate no new warnings
• I have checked my code and corrected any misspellings

Screenshots/Gifs

N/A

Further comments

None