Skip to content
Other

MiniManjoro-25M

крошечная RWKV-7 для русского чата на CPU

1 min read

MiniManjoro-25M — текстовая модель на 24.91M параметров от SatoshiManjoro. Она построена на архитектуре RWKV-7, заявлены языки ru/en, а весь цикл — от токенизации до обучения — рассчитан на обычный 4-ядерный x86 CPU.

Ключевые особенности

  • Архитектура RWKV-7: линейная рекуррентная динамика WKV-7 вместо квадратичного attention.
  • Ориентирована на x86 CPU: обучение и инференс в FP32, 4 потока.
  • На инференсе не нужен KV-кэш: модель хранит фиксированное состояние, память не растёт с длиной диалога.
  • Компактный Byte-level BPE-словарь на 8192 токена; матрица эмбеддингов — около 4.2M параметров.
  • Есть чекпоинт PyTorch и экспорт в GGUF для llama.cpp / C++ рантаймов.
  • Лицензия Apache 2.0.

Пример

pip install torch numpy tokenizers datasets zstandard huggingface_hub gguf
python download.py
python identity.py
python train.py
python chat.py
ПодробнееПараметры: 24.91M, 5 слоёв, hidden 512, 8 голов внимания, словарь 8192.
Обучение: корпус ~25 МБ, окно 128 токенов, батч 16, AdamW, 3500 шагов.
Инференс: CPU, 4 потока; состояние фиксированное, KV-кэш не используется.
Ограничения: крошечный корпус и словарь; бенчмарков в карточке нет.
Discuss Telegram 0 47

Will it run?

Runs in GGUF

The most precise of the ready quantised files fits.

  • GGUFweights 0.1 GB · ready files0.6 GB

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: GGUF — 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

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    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.

    # SatoshiManjoro/MiniManjoro-25M
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download SatoshiManjoro/MiniManjoro-25M --local-dir ~/models/MiniManjoro-25M
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to MiniManjoro-25M
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

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.

Model family

New posts about versions of MiniManjoro-25M (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

SatoshiManjoro/MiniManjoro-25M
Author
SatoshiManjoro
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

470

Ask AI about this model

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

Comments