Skip to content

Engine Inference Core E3: qwen3_5_moe, lfm2, nemotron_h #106

Description

@tlkahn

E3 - qwen3_5_moe, lfm2, nemotron_h

Remaining non-dense generate families: MoE+linear-attn (qwen3_5_moe) and hybrid conv/SSM (lfm2, nemotron_h).

Spec: doc/plans/engine-inference-core.md § E3

Part of: #56
Depends on: E1 (#104) for qwen3_5_moe MoE block; hybrid cache is E3-local (#117)

Tracking notes folded into sub-issues

Sub-issues

Suggested order

  1. E3: Hybrid cache (SSM/conv) + unified hybrid support gate #117 hybrid cache + gate unification
  2. E3: qwen3_5_moe (linear attention + MoE FFN) #118 / E3: lfm2 (short-conv + attention hybrid) #119 / E3: nemotron_h (mamba-style SSM + attention/MLP) #120 family forwards (parallelizable after E3: Hybrid cache (SSM/conv) + unified hybrid support gate #117; E3: qwen3_5_moe (linear attention + MoE FFN) #118 also needs E1)
  3. E3: Parity gates (qwen3_5_moe, lfm2, nemotron_h) #121 parity as each family lands

Acceptance

  • Each family generates on a canonical small checkpoint (skip-if-absent)
  • Hybrid cache correctly hosts SSM/conv entries alongside KV entries
  • Pure-dense rejects no longer use the Stage D triple-gate once hybrid is first-class

Metadata

Metadata

Assignees

Labels

engine-inferenceEngine inference core (issue #51) feature areaenhancementNew feature or request

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions