Skip to content

Conversation

@wqerrewetw
Copy link
Owner

Make sure to read the contributing guidelines before submitting a PR

wqerrewetw and others added 4 commits October 29, 2025 15:39
* Fix cl (#7)

* Rename build-amd.yml to build-amd.yml.disabled

* Rename winget.yml to winget.yml.disabled

* Rename server.yml to server.yml.disabled

* Rename build.yml to build.yml.disabled

* Update release.yml

* Rename build-cmake-pkg.yml to build-cmake-pkg.yml.disabled

* Rename build-linux-cross.yml to build-linux-cross.yml.disabled

* Rename build-riscv-native.yml.disabled to build-riscv-native.yml

* Rename docker.yml.disabled to docker.yml

* Rename update-ops-docs.yml to update-ops-docs.yml.disabled

* Remove macOS-arm64 job from release workflow

Removed macOS-arm64 job and its associated steps from the release workflow.
* CUDA: Fix bug in topk-moe for gpt-oss

When using ggml_can_fuse_subgraph, the output nodes which are passed are wrong. This causes `test-backend-ops` to still fuse ndoes (because the nodes are not used elsewhere in the graph),
but it actually doesn't fuse in the actual gpt-oss

* fix for qwen3 too

* change ifndef to ifdef
…rg#16793)

This lets the copy to the destination device use the host-visible
vidmem optimization.
@wqerrewetw wqerrewetw merged commit 2ea90c3 into qw25vl Oct 29, 2025
2 of 6 checks passed
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.

4 participants