[Anish Mehta (anishmehta24)]

Why are these changes needed?

add_cost_to_space loops over the keys of the search space, but when a categorical has a list low-cost point (one low-cost config per choice) it hits a return inside the loop. Every key after it is skipped: no low_cost_point, no bounded, and nested choices never get their const. The function also returns None instead of the constants.

With a second nested choice after the first one, tune.run crashes:

space = {
    "c": tune.choice([{"a": tune.randint(1, 10)}, {"b": tune.uniform(0, 1)}]),
    "x": tune.randint(1, 100),
    "y": tune.choice([{"p": tune.uniform(0, 1)}, {"q": tune.uniform(0, 1)}]),
}
tune.run(lambda c: {"metric": c["x"]}, config=space,
         low_cost_partial_config={"c": [{"a": 1}, {"b": 0}], "x": 1},
         metric="metric", mode="min", num_samples=10)
# AttributeError: 'Categorical' object has no attribute 'const'

Changing the return to continue keeps what happens for that key and lets the loop finish.

Related issue number

None found.

Checks

• I've used pre-commit to lint the changes in this PR (note the same in integrated in our CI checks).
• I've included any doc changes needed for https://microsoft.github.io/FLAML/. See https://microsoft.github.io/FLAML/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

test_add_cost_to_space_keys_after_nested_choice fails on main and passes with the change.

[Copilot]

🟢 Approval recommended

The localized control-flow fix addresses the reported failure and includes focused regression coverage.

No open issues 🎉

What changed in this PR

Fixes premature termination in add_cost_to_space so search-space initialization continues after a categorical list low-cost point.

Changes:

• Replaces return with continue to process remaining keys and return constants.
• Adds regression coverage for initialization and successful tune.run execution.

File	Description
test/​tune/​test_space.py	Tests subsequent keys, returned constants, and tuning execution.
flaml/​tune/​space.py	Continues initialization after handling a list low-cost point.

_{Review effort: Balanced}

[Li Jiang (thinkall)]

Overall review of the complete PR: no blocking code issue found; approval withheld pending CI.

Replacing the early return with continue preserves list-valued categorical handling while allowing later domains to be initialized and the constant dictionary to be returned. The focused test module passed all 3 tests locally, including the added tuning regression. The review-code harness is clean and Copilot's current Balanced review recommends approval with no open issues. Build checks on the current head are failing, so it does not satisfy the all-CI-passed approval gate. This is a formal comment review, not an approval.

Posted by thinkall-agent-auto-reviewer

[Copilot]

🟢 Approval recommended

The narrow control-flow fix addresses the reported failure with targeted regression coverage and no unresolved findings.

No open issues 🎉

_{Review effort: Balanced}

[Copilot]

🟢 Approval recommended

The localized control-flow fix has targeted regression coverage and no unresolved findings.

No open issues 🎉

_{Review effort: Balanced}

[Li Jiang (thinkall)]

Overall review at 7184aca9c957e58fe2176a51579f1e454cbaa122: re-reviewed the complete PR with review-code, including recursive categorical initialization and downstream consumers. Continuing the outer loop after handling a list low-cost point preserves that category's handling while allowing later constants, bounds, and nested choices to be initialized and the constants dictionary to be returned. No remaining correctness issue found.

Validation: python -m pytest test\tune\test_space.py -q --disable-warnings passed (3 passed). The new regression checks suffix constants, low-cost metadata, boundedness, subsequent nested-category initialization, and ten completed tuning trials. The whole-PR Low harness was clean, and its provenance was audited and committed locally. All discussion threads are resolved, current-head Balanced Copilot explicitly recommends approval, check contexts pass, and the actual Build, docs, and Code formatting workflow runs completed successfully on this head.

Posted by thinkall-agent-auto-reviewer