Skip to content
This repository was archived by the owner on Aug 4, 2026. It is now read-only.
This repository was archived by the owner on Aug 4, 2026. It is now read-only.

Showcase / question: a board-proven offline language runtime on ESP32-C3, and whether some lightweight language capability may eventually crystallize beyond dense models #98

Description

@Juemian-Deng

Hi Gemma folks,

I wanted to share a small but unusual language-runtime project that may still be relevant to the broader question of how far lightweight language capability can be pushed, even though it sits far outside the normal CPU/GPU/TPU dense-model path.

We built a public demo line called Engram and deployed it on a commodity ESP32-C3.

Current public numbers:

  • Host-side benchmark capability

    • LogiQA = 0.392523
    • IFEval = 0.780037
  • Published board proof

    • LogiQA 642 = 249 / 642 = 0.3878504672897196
    • host_full_match = 642 / 642
    • runtime artifact size = 1,380,771 bytes

Important scope note:

This is not presented as unrestricted open-input native LLM generation on MCU.

The board-side path is closer to a flash-resident, table-driven runtime with:

  • packed token weights
  • hashed lookup structures
  • fixed compiled probe batches
  • streaming fold / checksum style execution over precompiled structures

So this is not a standard lightweight dense model running on a smaller device. It is closer to a task-specialized language runtime whose behavior has been crystallized into a compact executable form under severe physical constraints.

Repo:
https://github.com/Alpha-Guardian/Engram

Why I’m posting here is that Gemma seems to represent one of the clearest public efforts around lightweight open language models with practical inference paths.

What I’d be curious about is whether systems like this should be thought of as:

  • completely outside the lightweight language-model family
  • an extreme endpoint where some task capability is no longer best served by even a compact dense model
  • or an early sign that future language systems may combine lightweight dense models with highly specialized executable forms for certain capability slices

If this direction is relevant to your team, I’d be glad to compare notes.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions