Skip to content
Language models

Ark-Cortex v3 GIRO Copilot

1.5B с нативными троичными весами

2 min readupdated

Ark-Cortex v3 (ACX-03) — модель генерации текста на 1.5B параметров от Arkheion Labs / CICLOGIRO для GIRO Desktop и экосистемы ArkheionNet. Основана на Qwen2/Qwen3 Causal LM, работает с нативным троичным представлением 1.58-bit (веса −1, 0, +1) и предназначена для локального инференса на кассах, локальных серверах и маломощных десктопах. Заявленные языки — португальский и английский.

Ключевые особенности

  • Нативная троичная квантизация: веса 1.58-bit со значениями −1, 0, +1 и группировкой масштабов g=64.
  • Два формата: .ark на 1.22 GB с zero-copy paged mmap и GGUF Q2_0 на 901 MB для llama.cpp и Ollama.
  • Архитектура Qwen 1.5B: 28 слоёв, hidden size 1536, GQA с 12 Q-головами и 2 KV-головами, head dim 128.
  • Заточка под edge: терминалы точек продаж, локальные серверы и десктопы с низким потреблением памяти.
  • Обучение: 5 600 локальных шагов с Multi-Token Prediction (MTP-4), финальный loss 2.13 (CE/KL).
  • Словарь на 151 936 BPE-токенов; в карточке указаны языки pt и en.

Пример

./llama-server -m acx_giro_1.5b_q2_0.gguf -c 2048 --port 8080 -ngl 99
ПодробнееФинальный loss — 2.13 (CE/KL) после 5 600 шагов обучения с MTP-4.
Размеры: .ark — 1.22 GB, GGUF Q2_0 — 901 MB; ориентир на низкое потребление памяти.
Публичных бенчмарков качества в карточке нет; заявленные языки — pt и en.
Для GGUF указаны llama.cpp/llama-server и Ollama; .ark рассчитан на GIRO Desktop.
Discuss Telegram 1 76

Will it run?

Runs in GGUF Q2_0

The most precise of the ready quantised files fits.

  • GGUF Q2_0weights 0.9 GB · ready files1.6 GB

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: GGUF Q2_0 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

A ready script for the machine we advise above: paste it and the model downloads and starts.

  1. 1. Rent A30

    $0.26 an hour with the coupon. Pick the Ubuntu image when you create the machine.

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Ooriginador/acx03-qwen3-1.5b-ternary-giro — A30, GGUF Q2_0
    # NVIDIA driver: CUDA 13 needs 580+. If older or missing, the current nvidia-open is installed (keep yours: KEEP_DRIVER=1)
    DRV=$(nvidia-smi --query-gpu=driver_version --format=csv,noheader 2>/dev/null | head -1 | cut -d. -f1)
    case "$DRV" in ""|*[!0-9]*) DRV=0 ;; esac
    if [ "$DRV" -lt 580 ] && [ -z "${KEEP_DRIVER:-}" ]; then
      . /etc/os-release && \
      curl -fsSLO "https://developer.download.nvidia.com/compute/cuda/repos/ubuntu${VERSION_ID//./}/x86_64/cuda-keyring_1.1-1_all.deb" && \
      sudo dpkg -i cuda-keyring_1.1-1_all.deb && rm -f cuda-keyring_1.1-1_all.deb && \
      sudo apt-get update && \
      sudo DEBIAN_FRONTEND=noninteractive apt-get install -y nvidia-open || \
      { echo "Could not install the NVIDIA driver: see the error above. To keep the current one, run with KEEP_DRIVER=1."; exit 1; }
      echo "NVIDIA driver updated. Reboot the machine (sudo reboot) and run this same script again."
      exit 0
    fi
    curl -fsSL https://ollama.com/install.sh | sh
    ollama run hf.co/Ooriginador/acx03-qwen3-1.5b-ternary-giro:Q2_0
  3. 3. Use it

    The chat opens right in the terminal. Leave with /bye, start again with the same ollama run command.

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

Model family

New posts about versions of acx03-qwen3-1.5b-ternary-giro (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

Ooriginador/acx03-qwen3-1.5b-ternary-giro
Author
Ooriginador
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

761

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments