Skip to content

feat(example): support chained NextN server MTP#2319

Merged
abetlen merged 1 commit into
mainfrom
feat/server-mtp-chain-heads
Jun 25, 2026
Merged

feat(example): support chained NextN server MTP#2319
abetlen merged 1 commit into
mainfrom
feat/server-mtp-chain-heads

Conversation

@abetlen

@abetlen abetlen commented Jun 23, 2026

Copy link
Copy Markdown
Owner

Adds server example support for chained NextN MTP draft models.

  • Detects multi-layer NextN draft models with llama_model_n_layer_nextn.
  • Uses llama_set_nextn_layer_offset while processing and drafting chained heads.
  • Keeps the sampled-batch MTP fast path limited to existing single-head draft models.

@abetlen abetlen merged commit 4ff48f0 into main Jun 25, 2026
15 checks passed
@abetlen abetlen deleted the feat/server-mtp-chain-heads branch June 25, 2026 06:34
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