Repository navigation
Conversation
EngiBench #280 and #281 rename the volume target to volfrac in thermoelastic2d and heatconduction2d and move both to new dataset repos holding the same rows. The committed specs named the old condition and dataset, so they no longer resolved. - Freeze thermoelastic2d/v3 and heatconduction2d/v3. Each is its v2 with the version, volume condition, dataset, digest and notes changed. Every metric, tolerance and seed is the same. - Delete the old versions of those two specs. Git history keeps them. - With no version given, load the newest spec committed for the problem, sorted numerically. This replaces a hardcoded v2 here and v1 in the CLI. - Pin CI to the EngiBench commit that has both renames, and assert both new dataset ids. - Rename the old condition names in test fixtures and the contributor guide. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #79. Merge #79 first, and this PR then retargets to
main. It depends on EngiBench IDEALLab/EngiBench#280 and IDEALLab/EngiBench#281.Why
EngiBench now names every volume target
volfrac. Thermoelastic2D wasvolume_fraction_targetand HeatConduction2D wasvolume. Both problems moved to new dataset repos holding the same rows under the new column name. The committed specs for these two problems named the old condition and the old dataset, so EngiOpt refused to resolve them.What changes
thermoelastic2d/v3andheatconduction2d/v3. Each is its v2 with only the version, the volume condition, the dataset, the digest and the notes changed. Every metric, tolerance, seed and sample count is the same. They were frozen on Euler against the EngiBench commit that has both renames, and the freeze script failed on any other field changing. The digest changes because it covers condition names as well as values.main.volfrac.Published checkpoints
340 heat conduction and 3 thermoelastic2d packages on HF record the old names in
metadata.json, incondition_keys. That list maps dataset columns to the network's inputs, so EngiOpt refuses these packages until it's updated. A script renames that one field in place, with one commit per model repo. It runs right after this PR merges, so no working setup sees a mismatch.Checks
Pre-commit passes. The offline spec, metrics, schema, checkpoint and evaluation tests pass: 107 passed, 6 deselected. The deselected tests need the network and run in CI.
🤖 Generated with Claude Code