Skip to content

feat(gallery): add Qwen3.8 9B Distill Q5 and Q6 - #12042

Open
localai-org-maint-bot wants to merge 1 commit into
masterfrom
cron/model-gallery-20260914-run4
Open

localai-org-maint-bot wants to merge 1 commit into
masterfrom
cron/model-gallery-20260914-run4

Conversation

@localai-org-maint-bot

Copy link
Copy Markdown
Collaborator

Description

Add Q5_K_M and Q6_K GGUF variants to the existing Qwen3.8 9B Distill gallery entry, filling the gap between Q4 and Q8. Both use the existing llama.cpp configuration and embedded chat template. Add CLI selection examples to the gallery documentation.

Notes for Reviewers

  • Source: https://huggingface.co/empero-ai/Qwen3.8-9B-Distill-GGUF
  • New downloads are pinned to revision 760121cd70bb4c36b2b5ec58eb765e0df5987efe.
  • Both SHA256 values agree across HF API LFS metadata, resolve-header x-linked-etag, and the repository's SHA256SUMS.
  • LocalAI's pinned llama.cpp revision df03399b885831b2a1603b3abb0d8c156808e363 registers qwen35. No backend change is needed.
  • Validation: a standalone Go yaml.v3 validator parsed all 1,849 entries and passed unique-name, variant-reference, nesting, single-owner, download-path, revision, SHA, configuration-parity, and unchanged-unrelated-entry checks. git diff --check passed.
  • go test ./core/gallery could not start because generated pkg/grpc/proto files are missing. make protogen-go is unavailable in this environment. CI should run the full gallery suite; no inference test was run.

Signed commits

  • Human DCO sign-off required; no AI sign-off added, per repository policy.
  • Documentation updated (docs/content/) for user-facing changes.

Fill the quantization gap between the existing Q4 and Q8 builds.
Pin the new downloads to a Hugging Face revision and retain the
existing llama.cpp chat configuration. Document explicit selection.

Assisted-by: Codex:GPT-6
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant