Skip to content

Bump Laya: source 7a77439b, model 7b928d82 - #568

Merged
bong-water-water-bong merged 1 commit into
mainfrom
bump-laya/7a77439b-7b928d82
Oct 25, 2026
Merged

bong-water-water-bong merged 1 commit into
mainfrom
bump-laya/7a77439b-7b928d82

Conversation

@bong-water-water-bong

Copy link
Copy Markdown
Collaborator

third_party/laya: 0a05a45052e2 -> 7a77439b0923 (NandhaKishorM/laya main). config/laya.json revision: 7b928d828b7b -> 7b928d828b7b (convaiinnovations/laya).

Before merging: scripts/fetch-laya.sh

(verifies every file) and the Laya tests.

@github-actions

Copy link
Copy Markdown

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 1 🔵⚪⚪⚪⚪
🧪 No relevant tests
🔒 No security concerns identified
⚡ No major issues detected

@bong-water-water-bong
bong-water-water-bong merged commit 982dd52 into main Oct 25, 2026
8 checks passed
@bong-water-water-bong
bong-water-water-bong deleted the bump-laya/7a77439b-7b928d82 branch October 25, 2026 12:57
@bong-water-water-bong

Copy link
Copy Markdown
Collaborator Author

Merged on the same reasoning as #538, after checking the delta: 0a05a45052e2 -> 7a77439b0923 is 86 commits /
72 files
, and every one of them is in the upstream SDK/reference trees that the engine does not consume
(laya/ Python, laya-java/, laya-rust/, laya-ts/, laya-dotnet/, sdk/, nix/, docs) — grep -rn "third_party/laya" over cmake/, CMakeLists.txt, scripts/, app/, tools/ finds nothing, and the tree builds
with third_party/laya empty. The router's scorer is third_party/ggmlc and the checkpoint comes from
~/models/laya-pinned.

The pin that matters, the model revision in config/laya.json (7b928d828b7b), is unchanged by this PR;
scripts/fetch-laya.sh verifies every file of it, and tests/laya_route_e2e.sh passes on the candidate build. CI green
(build, CodeQL, pr_agent); hw_gate skipped for a pin-only change.

bong-water-water-bong added a commit that referenced this pull request Oct 28, 2026
One line: `third_party/laya` `7a77439b0923` -> `b9fe351ce81e`. `config/laya.json` already pins model revision `7b928d828b7b` (unchanged since #317), so this is the source-only half of a bump — the sixth in a row (#439, #452, #459, #511, #538, #568) and the model the router scores is untouched. Nothing in the tree reads or compiles the submodule: `grep -rn third_party/laya` matches only `.github/workflows/bump-laya.yml`, the router builds engine-local `laya/` (`CMakeLists.txt:334`) and the optional fast scorer builds `third_party/ggmlc` (`cmake/laya-ggml.cmake:24`), so the bump-laya workflow's 'run scripts/fetch-laya.sh and the Laya tests' step has no code path to exercise and its HF fetch was deliberately not run while the measurement pass owns this box. Risk: a stale provenance reference only — worst case whoever next diffs the submodule against NandhaKishorM/laya main sees one more commit than the record, revertible by moving the one gitlink back.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant