ZIm/crates/eval/examples/checkpoint_stability/diff_criteria.md
Nathan Sobo bab28560ef
Systematically optimize agentic editing performance (#28961)
Now that we've established a proper eval in tree, this PR is reboots of
our agent loop back to a set of minimal tools and simpler prompts. We
should aim to get this branch feeling subjectively competitive with
what's on main and then merge it, and build from there.

Let's invest in our eval and use it to drive better performance of the
agent loop. How you can help: Pick an example, and then make the outcome
faster or better. It's fine to even use your own subjective judgment, as
our evaluation criteria likely need tuning as well at this point. Focus
on making the agent work better in your own subjective experience first.
Let's focus on simple/practical improvements to make this thing work
better, then determine how we can craft our judgment criteria to lock
those improvements in.

Release Notes:

- N/A

---------

Co-authored-by: Max <max@zed.dev>
Co-authored-by: Antonio <antonio@zed.dev>
Co-authored-by: Agus <agus@zed.dev>
Co-authored-by: Richard <richard@zed.dev>
Co-authored-by: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Michael Sloan <mgsloan@gmail.com>
2025-04-19 02:47:59 +00:00

1 KiB
Raw Blame History

  1. A new changeset file is created to document a patch that improves diff editing animations and enhances prompts for large file edits. An indicator showing the number of diff edits is also added next to each file path.
  2. In diff.ts, the error message thrown when a SEARCH block doesnt match content has been updated to clarify that the mismatch could be due to out-of-order blocks.
  3. In responses.ts, the assistant response for diff mismatches now recommends limiting to 13 SEARCH/REPLACE blocks at a time for large files. It also simplifies fallback instructions for using the write_to_file tool.
  4. The DiffViewProvider.ts file has been updated to replace line-by-line animations with chunk-based updates for better performance. For large diffs, a smooth scrolling animation is introduced to maintain visual context. Small diffs still scroll directly.
  5. In CodeAccordian.tsx, a new visual indicator displays the number of REPLACE blocks in the code diff using a diff icon and count, providing quick insight into the volume of changes.