README.md

# Digitable

We build **Digit**, a local agent for the Russian-language course platform
[courses.digitable.life](https://courses.digitable.life). Its design constraint is
unusual: the language model is not allowed to be a source of facts. Every fact in an
answer comes from a deterministic utility, a verbatim quote from the course corpus, or a
formal certificate — the model only chooses which of those to invoke.

The models published here are the piece that does the choosing.

## What is in this organisation

**These are tool routers, not general-purpose assistants.** Each one maps a user query to
a tool category, then emits a tool call with extracted arguments — or refuses. It does not
write the answer. Load one as a chat model and you will get nonsense, and none of our
published metrics describe that use.

| Repository | What it is |
|---|---|
| [`digit-router-0.6b`](https://huggingface.co/digitable-lol/digit-router-0.6b) | The shipping router. LoRA adapters (v1/v2/v3) over `Qwen/Qwen3-0.6B` plus merged GGUF quantisations. 424 MiB at Q5_K_M, ~500 ms per full two-step routing cycle on 8 CPU threads. |
| [`digit-router-1.7b`](https://huggingface.co/digitable-lol/digit-router-1.7b) | The same training run at 1.7B. Higher routing accuracy; measurably not worth 3× the parameters for this task. |
| [`digit-router-experiments`](https://huggingface.co/digitable-lol/digit-router-experiments) | Three adapters that lost — Vikhr, ruadapt, QVikhr-3. Published so the negative result stays reproducible instead of becoming folklore. |

## How to read our numbers

Two conventions run through every model card, and both exist because the obvious way to
report these numbers is misleading:

* **A counted refusal is not a conscious refusal.** An eval harness scores an unparseable
  answer as a refusal, so a model that merely breaks scores like a model that knows when
  to decline. We always report both columns. The untuned 0.6B base scores 75.3 % counted
  against 9.3 % conscious — a 66-point gap that is entirely broken output.
* **Known defects are stated before the good tables, not in a footnote.** The imatrix
  quantisation of the 0.6B model measurably breaks its ability to refuse and carries a
  do-not-deploy warning next to the file. At Q4 the routers do not emit garbage; they emit
  structurally flawless tool calls with invented arguments, and a GBNF grammar does not
  catch that. Our v3 adapters are a routing regression against v2 on a single seed. All of
  this is on the model pages.

Every published file's sha256 is recorded in a `MANIFEST.json` in its repository. We track
runs by weight hash rather than by tag, because a tag was once re-created from a different
build while a 250-task evaluation was in flight.

Base models are `Qwen/Qwen3-*` under Apache-2.0. The training data is derived from a
GPL-3.0 utility catalogue; we state that provenance on every page and do not claim to have
resolved what it means for weights.

---

## По-русски

Мы делаем **Digit** — локального агента для платформы курсов
[courses.digitable.life](https://courses.digitable.life). Ограничение архитектуры
необычное: языковой модели запрещено быть источником фактов. Содержание ответа даёт
детерминированная утилита, дословная цитата из корпуса курсов или формальный сертификат.
Модель выбирает, что вызвать, — и только.

**Здесь лежат маршрутизаторы, а не универсальные ассистенты.** Модель относит запрос к
категории инструментов и извлекает аргументы либо отказывается; ответ она не пишет. Если
загрузить её как чат-модель, вы получите бессмыслицу, и опубликованные метрики к такому
использованию не относятся.

Два правила чтения наших чисел. Первое: **засчитанный отказ ≠ осознанный** — харнесс
считает отказом любой неразбираемый ответ, поэтому сломанная модель выглядит как
осторожная; мы всегда печатаем обе колонки. Второе: **известные дефекты стоят до таблиц с
хорошими числами, а не в примечаниях.** imatrix-квант 0.6B ломает способность отказываться
и помечен как непригодный к поставке; при Q4 модель выдаёт структурно безупречные вызовы с
выдуманными аргументами, и грамматика этого не ловит; адаптеры v3 — измеренный регресс
маршрутизации против v2 на одном seed.

sha256 каждого опубликованного файла записан в `MANIFEST.json` соответствующего
репозитория.