Skip to content
Other

SiliconShield Muse Glimmer RTL

LoRA-адаптер для проверки RTL по требованиям

2 min read

SiliconShield адаптирует Muse Glimmer 30B, чтобы оценивать переданный RTL-регион на соответствие английскому требованию и явным допущениям выполнения. Это LoRA-адаптер примерно на 183 MiB для Meta Muse Glimmer 30B: исходная модель обязательна, отдельной или merged-моделью адаптер не является. Один адаптер покрывает локальную защиту регистров и отдельные контракты для FIFO, счётчиков и арбитража.

Ключевые особенности

  • Возвращает структурированные оценки для satisfies_requirement, violates_requirement и insufficient_context.
  • Решение относится к заданному требованию и допущениям, а не ко всем возможным свойствам дизайна.
  • Обучение: 47 923 200 LoRA-параметров, rank 8, alpha 16, dropout 0; цель — next-token cross-entropy по одному токену ответа A/B/C.
  • Заявленный путь запуска — BF16 на трёх CUDA-видеокартах; другие варианты размещения и квантизация не квалифицированы.
  • Вход длиннее 2048 токенов отклоняется, а не обрезается.

Пример

CUDA_VISIBLE_DEVICES=0,1,2 python predict.py --design design.sv --requirement "Outside reset, cfg_reg may change only when privileged is asserted." --assumptions "Two-state synchronous logic. Inputs are defined and may vary independently. Reset is exempt from the requirement."
ПодробнееНа development-популяциях: local guards — 280/288 (97.22%), focused temporal — 223/234 (95.30%), combined — 503/522 (96.36%).
Нужны оригинальные веса Muse Glimmer 30B и три NVIDIA RTX PRO 4000 Blackwell примерно по 24 GiB; это не независимый sealed-тест, а повторно использованные development-данные.
Не заявлены whole-chip correctness, произвольное multi-cycle reasoning, CDC, криптографическая безопасность, physical-fault analysis и поиск ошибок на уровне репозитория.
Discuss Telegram 0 47

Will it run?

Runs in 4-bit

Full precision does not fit, a quantised version does: quality barely changes down to 8 bits, more noticeably at 4 bits.

  • BF16weights 56 GB · ready files62 GB
  • FP8 / INT8recommendedweights 28 GB31 GB
  • 4-bitweights 16 GB18 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

This is a LoRA: counted together with its base model Muse-Glimmer-30B.

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.

    # debjyotikm/siliconshield-muse-glimmer-rtl-lora
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    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 meta-models/Muse-Glimmer-30B --local-dir ~/models/Muse-Glimmer-30B
    hf download debjyotikm/siliconshield-muse-glimmer-rtl-lora --local-dir ~/models/siliconshield-muse-glimmer-rtl-lora
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Muse-Glimmer-30B
  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

Requires the base model

meta-models/Muse-Glimmer-30B

This model is a LoRA / adapter. Download the base model first, then load this one on top.

New posts about versions of Muse-Glimmer-30B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 1All on Hugging Face
Merges 1All on Hugging Face

About the model

debjyotikm/siliconshield-muse-glimmer-rtl-lora
Author
debjyotikm
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