[Docs] Fix doc quality issues on init_options.md - #869
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0bc47786f4
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| | `debug=True` | on | on | | ||
|
|
||
| - `debug=True` always implies `check_out_of_bound=True` (the field-bounds check fires whenever debug mode is on). | ||
| - `debug=True` always implies `check_out_of_bound=True` (the field-bounds check fires whenever debug mode is on), except on Metal and Vulkan, where the field-bounds check is unavailable and is reset to `False` regardless of `debug` (see the per-backend table below). |
There was a problem hiding this comment.
For Metal and Vulkan users, this correction now conflicts with the linked docs/source/user_guide/debug.md, which still says both that debug=True makes out-of-bounds accesses raise and that it “always implies” check_out_of_bound=True (lines 15 and 68). Update that user-facing guide as well so users do not receive contradictory guidance depending on which debug documentation page they read.
Useful? React with 👍 / 👎.
Move the three-layer offline_cache internals (KernelCompilationManager, PtxCache, PTX/LLVM-IR/SASS, ptxas/cuModuleLoadDataEx, driver compute cache, per-process nonce) into a clearly delimited "Under the hood (advanced)" subsection, and keep a concise user-facing offline_cache description. De-jargon the debug section (drop "IR verification / verifier walks the IR / compiler pass" wording). Resolves RULE 1 (undefined terms) and RULE 2 (internal material outside an advanced section) flagged by .github/workflows/check_doc_quality.yml.
7b30a10 to
43a911d
Compare
![graphite-app[bot]](https://avatars.githubusercontent.com/in/158384?s=60&v=4){kind=small}
|
|
||
| Enables: | ||
| - field-bounds check on tensor indexing (out-of-range index raises `RuntimeError`); | ||
| - kernel `assert` statements; |
There was a problem hiding this comment.
The PR description claims to fix the autodiff link to use ./ and #what-can-go-wrong anchor, but this line shows no changes in the diff. The link remains as (autodiff.md) when it should be updated to (./autodiff.md#what-can-go-wrong) according to the stated fix.
-The adstack-overflow check on reverse-mode autodiff runs unconditionally on every backend regardless of `debug`; see [Autodiff -> What can go wrong](autodiff.md) for the contract.
+The adstack-overflow check on reverse-mode autodiff runs unconditionally on every backend regardless of `debug`; see [Autodiff -> What can go wrong](./autodiff.md#what-can-go-wrong) for the contract.Spotted by Graphite
Is this helpful? React 👍 or 👎 to let us know.
…tions - Rule 2 (scope): drop the internal CompileConfig struct name and the C++ header "source of truth" link from the user-facing intro. - Rule 1 (term): define "primal" inline at its first occurrence in the ad_stack_experimental_enabled section. - Gloss "control-flow-graph optimization" for cfg_optimization so no bare compiler jargon remains outside the advanced section. - Restore the autodiff link to ./autodiff.md#what-can-go-wrong (matches the PR description; flagged by the Graphite AI review).
1 participant
Issue: #
Brief Summary
copilot:summary
Walkthrough
copilot:walkthrough