Skip to content

Fix flaky reasoning budget tests on Metal by using greedy sampling #419

Fix flaky reasoning budget tests on Metal by using greedy sampling

Fix flaky reasoning budget tests on Metal by using greedy sampling #419

Triggered via pull request June 21, 2026 15:24
Status Success
Total duration 1h 5m 58s
Artifacts 17

publish.yml

on: pull_request
Start gate (abort window)
3s
Start gate (abort window)
Build WebUI assets (shared)
1m 17s
Build WebUI assets (shared)
Code style (spotless) + package graph
47s
Code style (spotless) + package graph
C++ Tests Ubuntu Latest x86_64
5m 47s
C++ Tests Ubuntu Latest x86_64
Test (vmlens interleavings)
1m 8s
Test (vmlens interleavings)
Cross-Compile manylinux_2_28 x86_64 (CUDA)
14m 51s
Cross-Compile manylinux_2_28 x86_64 (CUDA)
Cross-Compile Android aarch64 (OpenCL/Adreno)
1m 59s
Cross-Compile Android aarch64 (OpenCL/Adreno)
Build and Test macOS 15 arm64 (Metal)
2m 29s
Build and Test macOS 15 arm64 (Metal)
Build and Test Windows 2025 x86_64 (Ninja Multi-Config + sccache, eval)
4m 8s
Build and Test Windows 2025 x86_64 (Ninja Multi-Config + sccache, eval)
Build and Test Linux aarch64
2m 58s
Build and Test Linux aarch64
Cross-Compile Android aarch64
2m 6s
Cross-Compile Android aarch64
Build and Test Windows 2025 x86 (VS 2026)
16m 35s
Build and Test Windows 2025 x86 (VS 2026)
Build and Test Windows 2025 x86 (Ninja Multi-Config + sccache, eval)
3m 51s
Build and Test Windows 2025 x86 (Ninja Multi-Config + sccache, eval)
Cross-Compile manylinux2014 x86_64
1m 50s
Cross-Compile manylinux2014 x86_64
Build and Test macOS 14 arm64 (Metal)
2m 25s
Build and Test macOS 14 arm64 (Metal)
Build and Test macOS 15 arm64 (no Metal)
2m 35s
Build and Test macOS 15 arm64 (no Metal)
Build and Test Windows 2025 x86_64 (VS 2026)
16m 19s
Build and Test Windows 2025 x86_64 (VS 2026)
Java Tests macOS 15 arm64 (Metal)
22m 59s
Java Tests macOS 15 arm64 (Metal)
Java Tests Windows 2025 x86_64 (Ninja, eval)
17m 23s
Java Tests Windows 2025 x86_64 (Ninja, eval)
Java Tests Ubuntu Latest x86_64
13m 41s
Java Tests Ubuntu Latest x86_64
Java Tests macOS 14 arm64 (Metal)
18m 44s
Java Tests macOS 14 arm64 (Metal)
Java Tests macOS 15 arm64 (no Metal)
19m 58s
Java Tests macOS 15 arm64 (no Metal)
Java Tests Windows 2025 x86_64 (VS 2026)
15m 42s
Java Tests Windows 2025 x86_64 (VS 2026)
Check: v* tag
Check: v* tag
Check: main branch / SNAPSHOT
Check: main branch / SNAPSHOT
Publish Release to Central
Publish Release to Central
Publish Snapshot to Central
Publish Snapshot to Central
Attach Signed Binaries to GitHub Release
0s
Attach Signed Binaries to GitHub Release
Update Snapshot Pre-release on GitHub
0s
Update Snapshot Pre-release on GitHub
Fit to window
Zoom out
Zoom in

Deployment protection rules

Reviewers, timers, and other rules protecting deployments in this run
Event Environments Comment
Waiting
Wait timer
skipped by bernardladenthin Jun 21, 2026
startgate

Annotations

1 error, 6 warnings, and 3 notices
Java Tests Ubuntu Latest x86_64
/home/runner/work/java-llama/cpp/java-llama/cpp/target/classes/net/ladenthin/llama/value/CompletionResult
Build and Test macOS 14 arm64 (Metal)
The following taps are not trusted: aws/tap azure/bicep Homebrew will ignore formulae, casks and commands from these taps when `HOMEBREW_REQUIRE_TAP_TRUST` is set. This will become the default in Homebrew 6.0.0 or 5.2.0, whichever comes first. Enable trust checks now with: export HOMEBREW_REQUIRE_TAP_TRUST=1 Trust specific formulae, casks or commands with: brew trust --formula <user>/<tap>/<formula> brew trust --cask <user>/<tap>/<cask> brew trust --command <user>/<tap>/<command> or trust installed formulae from these taps with: brew trust --formula azure/bicep/bicep You can trust all formulae, casks and commands from these taps with: brew trust aws/tap azure/bicep Prefer trusting only the specific formulae, casks or commands you need. Untap them with: brew untap aws/tap azure/bicep To keep allowing them by default during the transition: export HOMEBREW_NO_REQUIRE_TAP_TRUST=1 This is not recommended and will be removed in a later release.
Build and Test macOS 15 arm64 (Metal)
The following taps are not trusted: aws/tap azure/bicep Homebrew will ignore formulae, casks and commands from these taps when `HOMEBREW_REQUIRE_TAP_TRUST` is set. This will become the default in Homebrew 6.0.0 or 5.2.0, whichever comes first. Enable trust checks now with: export HOMEBREW_REQUIRE_TAP_TRUST=1 Trust specific formulae, casks or commands with: brew trust --formula <user>/<tap>/<formula> brew trust --cask <user>/<tap>/<cask> brew trust --command <user>/<tap>/<command> or trust installed formulae from these taps with: brew trust --formula azure/bicep/bicep You can trust all formulae, casks and commands from these taps with: brew trust aws/tap azure/bicep Prefer trusting only the specific formulae, casks or commands you need. Untap them with: brew untap aws/tap azure/bicep To keep allowing them by default during the transition: export HOMEBREW_NO_REQUIRE_TAP_TRUST=1 This is not recommended and will be removed in a later release.
Build and Test macOS 15 arm64 (no Metal)
The following taps are not trusted: aws/tap azure/bicep Homebrew will ignore formulae, casks and commands from these taps when `HOMEBREW_REQUIRE_TAP_TRUST` is set. This will become the default in Homebrew 6.0.0 or 5.2.0, whichever comes first. Enable trust checks now with: export HOMEBREW_REQUIRE_TAP_TRUST=1 Trust specific formulae, casks or commands with: brew trust --formula <user>/<tap>/<formula> brew trust --cask <user>/<tap>/<cask> brew trust --command <user>/<tap>/<command> or trust installed formulae from these taps with: brew trust --formula azure/bicep/bicep You can trust all formulae, casks and commands from these taps with: brew trust aws/tap azure/bicep Prefer trusting only the specific formulae, casks or commands you need. Untap them with: brew untap aws/tap azure/bicep To keep allowing them by default during the transition: export HOMEBREW_NO_REQUIRE_TAP_TRUST=1 This is not recommended and will be removed in a later release.
Build and Test Windows 2025 x86 (Ninja Multi-Config + sccache, eval)
Node.js 20 is deprecated. The following actions target Node.js 20 but are being forced to run on Node.js 24: ilammy/msvc-dev-cmd@v1. For more information see: https://github.blog/changelog/2025-09-19-deprecation-of-node-20-on-github-actions-runners/
Build and Test Windows 2025 x86_64 (Ninja Multi-Config + sccache, eval)
Node.js 20 is deprecated. The following actions target Node.js 20 but are being forced to run on Node.js 24: ilammy/msvc-dev-cmd@v1. For more information see: https://github.blog/changelog/2025-09-19-deprecation-of-node-20-on-github-actions-runners/
Report
Node.js 20 is deprecated. The following actions target Node.js 20 but are being forced to run on Node.js 24: advanced-security/maven-dependency-submission-action@v5. For more information see: https://github.blog/changelog/2025-09-19-deprecation-of-node-20-on-github-actions-runners/
Report
Snapshot successfully created at 2026-06-21T16:30:41.613Z
Report
{ "manifests": { "llama": { "resolved": { "pkg:maven/org.projectlombok/lombok@1.18.46?type=jar": { "package_url": "pkg:maven/org.projectlombok/lombok@1.18.46?type=jar", "relationship": "direct", "scope": "runtime", "dependencies": [] }, "pkg:maven/org.junit.jupiter/junit-jupiter@6.1.0?type=jar": { "package_url": "pkg:maven/org.junit.jupiter/junit-jupiter@6.1.0?type=jar", "relationship": "direct", "scope": "development", "dependencies": [ "pkg:maven/org.junit.jupiter/junit-jupiter-api@6.1.0?type=jar", "pkg:maven/org.junit.jupiter/junit-jupiter-params@6.1.0?type=jar", "pkg:maven/org.junit.jupiter/junit-jupiter-engine@6.1.0?type=jar" ] }, "pkg:maven/org.junit.jupiter/junit-jupiter-api@6.1.0?type=jar": { "package_url": "pkg:maven/org.junit.jupiter/junit-jupiter-api@6.1.0?type=jar", "relationship": "indirect", "scope": "development", "dependencies": [ "pkg:maven/org.opentest4j/opentest4j@1.3.0?type=jar", "pkg:maven/org.junit.platform/junit-platform-commons@6.1.0?type=jar" ] }, "pkg:maven/org.opentest4j/opentest4j@1.3.0?type=jar": { "package_url": "pkg:maven/org.opentest4j/opentest4j@1.3.0?type=jar", "relationship": "indirect", "scope": "development", "dependencies": [] }, "pkg:maven/org.junit.platform/junit-platform-commons@6.1.0?type=jar": { "package_url": "pkg:maven/org.junit.platform/junit-platform-commons@6.1.0?type=jar", "relationship": "indirect", "scope": "development", "dependencies": [] }, "pkg:maven/org.junit.jupiter/junit-jupiter-params@6.1.0?type=jar": { "package_url": "pkg:maven/org.junit.jupiter/junit-jupiter-params@6.1.0?type=jar", "relationship": "indirect", "scope": "development", "dependencies": [] }, "pkg:maven/org.junit.jupiter/junit-jupiter-engine@6.1.0?type=jar": { "package_url": "pkg:maven/org.junit.jupiter/junit-jupiter-engine@6.1.0?type=jar", "relationship": "indirect", "scope": "development", "dependencies": [ "pkg:maven/org.junit.platform/junit-platform-engine@6.1.0?type=jar" ] }, "pkg:maven/org.junit.platform/junit-platform-engine@6.1.0?type=jar": { "package_url": "pkg:maven/org.junit.platform/junit-platform-engine@6.1.0?type=jar", "relationship": "indirect", "scope": "development", "dependencies": [] }, "pkg:maven/org.hamcrest/hamcrest@3.0?type=jar": { "package_url": "pkg:maven/org.hamcrest/hamcrest@3.0?type=jar", "relationship": "direct", "scope": "development", "dependencies": [] }, "pkg:maven/net.jqwik/jqwik@1.9.3?type=jar": { "package_url": "pkg:maven/net.jqwik/jqwik@1.9.3?type=jar", "relationship": "direct", "scope": "development", "dependencies": [ "pkg:maven/org.apiguardian/apiguardian-api@1.1.2?type=jar", "pkg:maven/net.jqwik/jqwik-api@1.9.3?type=jar", "pkg:maven/net.jqwik/jqwik-web@1.9.3?type=jar",
Report
Submitting snapshot...

Artifacts

Produced during runtime
Name Size Digest
Linux-Android-aarch64-libraries
6.81 MB
sha256:bd6b889ab5c51f982e34ca0645bb9a584417f7ea54da0844984ede421426a7c6
Linux-aarch64-libraries
7.23 MB
sha256:cd2f0fc7f8f14d737ae3b7a08ac71fb622ad9dba9887d91cead6cb85e09698e6
Linux-x86_64-libraries
7.46 MB
sha256:f43a94be08baa539251500c8bf867dffd92d14aad25aba5398f18a554d5c108d
Windows-x86-libraries
5.32 MB
sha256:0b459b55689aa883f6c5bbc4c0f3e7a2539a77ce49c700dd8de86e26945a7dd7
Windows-x86-ninja
5.34 MB
sha256:b8329e240083691e578fdc7e11c098bc9b6d2e013b0eaf79e951bf4fe59f16ab
Windows-x86_64-libraries
5.66 MB
sha256:c0429d10b4f819d183c124bf15941758f2b6701fa3834a8d135654e3ae9f72ad
Windows-x86_64-ninja
5.66 MB
sha256:d964706f4c2b8e68d3fb391fe2e89ed808afc7787151df1bda1bd3c2299b1b9a
android-libraries-opencl
7.18 MB
sha256:cfec03956a7ef28915ff2fdaccea4be7c8e8b0d0d5c4419e0dbe854ff568c182
jacoco-report
32.9 KB
sha256:bfe707f0d25da19550b8e46c0e83e0faeeba3a5d4f205c7643491ce4a176c885
linux-libraries-cuda
140 MB
sha256:e9851d4d19072f052a500ef96414fe22d5385354e30c475ce28cc523213d8009
llama-jars
278 MB
sha256:ce7b0078d7cee303b5391aa45e38fab6460aa5084346c1b99417745441b490aa
macos-14-libraries
6.67 MB
sha256:8fb5e5ca7ff887934bc8b6af6a9ccf7db8730e6f8ba79ebf3ac496fd564c9f00
macos-15-libraries
6.49 MB
sha256:94969fac750927ebe2a61985258a61836f82c35e5b4b44feb579a7bba4ba645a
macos-15-metal-libraries
6.49 MB
sha256:123e0c0eba94a84fde35d4a14c807aa814d86561ce32eeb5a293918f387e6bd2
pit-reports
126 KB
sha256:be046dcfa66758aaf0d03e636cfdd793faa7fdce0e90368ca7ba410f55ce891b
vmlens-report
190 KB
sha256:37072c8630af6547a6b854037ba620dfc92de46b9fc2a1330f0c70853019524b
webui-generated Expired
3.81 MB
sha256:589634dab247e09ca7c8cf74983e04e80718ba1af91825d7d259dc66cda8cff2