Skip to content

feat(gallery): add LFM2.5 8B DSpark variants - #11751

Open
localai-org-maint-bot wants to merge 1 commit into
masterfrom
cron/model-gallery-20260827-lfm8-dspark
Open

feat(gallery): add LFM2.5 8B DSpark variants#11751
localai-org-maint-bot wants to merge 1 commit into
masterfrom
cron/model-gallery-20260827-lfm8-dspark

Conversation

@localai-org-maint-bot

Copy link
Copy Markdown
Collaborator

Description

Adds LiquidAI LFM2.5-8B-A1B DSpark variants to the existing gallery family:

  • Q4_K_M target with a Q4_K_M draft sidecar
  • Q8_0 target with the recommended F16 draft sidecar

Both entries use llama.cpp mainline DSpark speculative decoding. The target model verifies every proposed token, so the variants preserve target output while improving generation speed. All four SHA256 values were verified from Hugging Face x-linked-etag headers.

Notes for Reviewers

Verification:

  • git diff --check
  • go test ./core/gallery -ginkgo.focus="gallery/index.yaml variant invariants" -ginkgo.v
  • go test ./core/gallery -ginkgo.focus="keeps every entry that declares variants installable" -ginkgo.v

The full go test ./core/gallery/... run reached 375 passing specs; three existing network-dependent specs failed because this environment returned 403 for raw GitHub and Gist URLs.

Signed commits

  • Yes, I signed my commits. AI maintainers cannot provide DCO sign-off.
  • Documentation updated (docs/content/) for user-facing changes, or not applicable. Gallery-only entry; not applicable.

Add Q4 and Q8 target builds with LiquidAI DSpark draft sidecars.
The variants expose mainline llama.cpp speculative decoding for the
existing LFM2.5 8B family.

Assisted-by: Codex:gpt-5
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