Dataset Usage & Eval Format Clarification #13

Hi, thanks for releasing LocAgent — great work!

A few quick questions:

1. Could you clarify how to use the fields in Loc-Bench_V1 when developing or evaluating new agents? A short description of each field and the label format would help a lot.

2. For evaluation, is there a canonical JSONL format for the results? The evaluation/run_evaluation.ipynb notebook is helpful, but a concrete example or format spec would be great — especially for batch evaluation runs.

3. Any plans for a leaderboard or a standard evaluation protocol for Loc-Bench?

Just want to make sure we're aligned with how the benchmark is intended to be used. Appreciate it!