Showcase / question: a board-proven offline language runtime on ESP32-C3, and whether some language capability may eventually diverge from dense tuning paths

[Alpha-Guardian]

Hi FMS HF Tuning folks,

I wanted to share a small but unusual language-runtime project that may still be relevant to the broader question of how language capability is tuned and developed, even though it sits far outside the usual GPU fine-tuning and pretraining 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 dense tuning or inference stack at all. 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 this repo represents one of the clearest public pathways for shaping model capability through tuning, adaptation, and
continued training.

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

• completely outside the normal tuning family
• an extreme endpoint where some task capability is no longer best shaped by denser tuning paths alone
• or an early sign that future language systems may combine dense tuning for broad capability with highly specialized executable forms for certain capability slices

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