Repository navigation
Tensor and model files onto the slot API (#1665 arm D B5) - #1713
Merged
Merged
Conversation
Batch B5 of the arm D step-2 design: builtins_tensor.c, model_infer.c, model_train.c and model_io.c leave the legacy borrow API and join the migrated-files gate (Codex Cloud builder). Orchestrator fixes from the local gates: the bool fuzz ([0fb]) found relu/leaky_relu SIGSEGV on a ragged tensor ([[7, 1], 1]) because tensor_to_flat dereferenced list_get_ref(v, r) -- NULL for a number by contract. Same class fixed in tensor_to_flat's EIGS_STRICT=0 path and in gather, and the two broadcast sites in tensor_elementwise's caller now use views so a number row stays a scalar (main's behaviour) instead of raising. [0fh] gains the three ragged cases. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KC99CmwatKssQYggkBCgwF
…_to_flat helper changed; no policy change) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KC99CmwatKssQYggkBCgwF
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
6 tasks
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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Part of #1665: batch B5 of the arm D step-2 design.
builtins_tensor.c,model_infer.c,model_train.candmodel_io.cleave the legacy borrow API and join the migrated-files gate. Storage staysValue**.Fixes from the local gates
Codex built the migration and ran only the static gates and a rehearsal smoke. The full local run found:
relu/leaky_reluSIGSEGV on a ragged tensor ([[7, 1], 1]), found by the bool fuzz[0fb].tensor_to_flatdereferencedlist_get_ref(v, r), which is NULL for a number by contract. The same class is fixed intensor_to_flat'sEIGS_STRICT=0path and ingather.list_get_ref(NULL) for a number row, soadd of [[[1, 2], 3], [1, 1]]raised where main broadcasts the scalar. They now use views, and the results match main.[0fh]gains the three ragged cases: 10/10 pass.tensor_to_flathelper; no policy changed.Gates (run locally, on top of B3)
make http(model files)jit-smokejit_diffdetect_leaks=1[99s]passing again, and the pre-existing[99d](#1708)[99d][70d](#1709)🤖 Generated with Claude Code
https://claude.ai/code/session_01KC99CmwatKssQYggkBCgwF