Downloads · 30 days
279
22% of all-time downloads
engram-md-demo
lewisdog/engram-md-demo
engram-md-demo is a machine learning model from lewisdog. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Engram cartridges are small, detachable n-gram fact memories (DeepSeek-style hash-keyed lookup tables) that mount onto a frozen base model in milliseconds, add exact-recall knowledge the base model cannot have, and un…
All-time downloads
1.3K
Public
Repo size
3.6 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.gguf3.3 GB · 100%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Aug 29, 2026
- Updated
- Sep 22, 2026
- SHA
- 304c9d39
Base models
Model card
From the Hugging Face model README
Engram cartridges — pluggable fact memory for a frozen model
Engram cartridges are small, detachable n-gram fact memories (DeepSeek-style hash-keyed lookup tables) that mount onto a frozen base model in milliseconds, add exact-recall knowledge the base model cannot have, and unmount leaving the base byte-identical. Built with aoa-engram and runnable in llama.cpp (engram branch) natively or in the browser via a wllama fork.
Files
| File | What it holds | Recall |
|---|---|---|
qwen3-0.6b-base-f16.gguf | The exact frozen base (F16 conversion of Qwen/Qwen3-0.6B-Base) all cartridges are fingerprint-bound to | — |
cartridge-premier-league-2526.gguf | Every 2025-26 Premier League result (380 matches) | 100% (full-store greedy-generation audit) |
cartridge-premier-league-2425.gguf | Every 2024-25 Premier League result (380 matches) | 100% (full-store greedy-generation audit) |
format-lora-2324.gguf | LoRA behaviour adapter: answer format + stopping, trained on the 2023-24 season | — |
cartridge-cities-16k.gguf | Populations of the world's 16,000 largest cities (GeoNames) | 99.49% (full-store deployment-exact audit; the 81 residual misses ship in stored-keys.json) |
qwen3-0.6b-chat-f16.gguf | Qwen3-0.6B (instruct) F16 — the chat workspace's base; its cartridges are bound to it | — |
cartridge-chat-premier-league-2526.gguf | 2025-26 results, rebuilt against the instruct base | 100% (full-store audit) |
cartridge-chat-premier-league-2425.gguf | 2024-25 results, rebuilt against the instruct base | 100% (full-store audit) |
cartridge-chat-cities-16k.gguf | 16,000 city populations, rebuilt against the instruct base | 99.48% (full-store audit; misses in stored-keys.json) |
engram.md | Machine-readable manifest: key formats, fact counts, audits, base-model fingerprints | — |
stored-keys.json | Every audited key per cartridge, plus each cartridge's known-miss list | — |
demo-facts.json | Sample prompts with ground-truth answers for the browser demo | — |
Both Premier League seasons finished after the base model's training cutoff — the frozen model provably cannot know these results. The two football cartridges are trained against the base merged with the format LoRA (a fact cartridge must be built against the stack it runs on); the cities cartridge is trained against the bare base.
Query format is exact (see engram.md), e.g.:
Newcastle v Liverpool | result | -> 2-3 (Liverpool)
Nanjing, CN | population | -> 9314685
Data sources & licenses
- Base model: Qwen/Qwen3-0.6B-Base, Apache-2.0.
- Football results: football-data.co.uk (free historical results data).
- City populations: GeoNames
cities15000, CC-BY 4.0.
- License
- apache-2.0
- Created
- Aug 29, 2026
- Updated
- Sep 22, 2026