Skip to content
Other

kibble-pythia-1b

1000 фактов в LLM без обучения

2 min read

Это набор из 1000 фактов и загрузчик, которые добавляют в EleutherAI/pythia-1b гейтированные нейроны рядом со слоем 6. Каждый факт хранится в отдельном слоте, а веса pythia не меняются: после удаления пака модель возвращается к исходному состоянию бит в бит. Автор — skyuu72; факты вымышленные, язык — английский.

Ключевые особенности

  • 1000 фактов о людях, компаниях, городах, таблетках и минералах, которых не существует.
  • Без обучения и без правки pretrained-весов: только forward passes и один linear solve.
  • Пак на 39 МБ: на факт адрес 8192-d и значение 2048-d; загрузчик и демо — в kibble.py.
  • По их замеру score 0.763 против MEMIT 0.625 и AlphaEdit 0.629; KL-дрейф 0.004, реальные факты сохранены 0.983.
  • Ограничения: опечатки в имени роняют exact match до 0.51; чужое совпадение по фамилии открывает слот в 71% случаев.

Пример

import kibble
model, tok = kibble.load()            # plain EleutherAI/pythia-1b
handle, meta = kibble.attach(model)   # 1000 facts in
print(kibble.say(model, tok, "Kryskorn Skeithhaes was born in the city of"))
handle.remove()                       # and out again
ПодробнееПак: 1000 слотов, 39 МБ; всё гонялось на RTX 4060 Laptop с лимитом 7 ГБ.
Методика: forward-only поиск значений, joint address solve, гейт out += clamp((W k - 0.35) / 0.2, 0, 1) @ V.
Потолок: факт в контексте с идеальным retriever даёт 0.871 — выше, чем запись в веса.
Адресный solve упирался в VRAM между 3800 и 4800 слотами; качество до этого не падало.
Примеры
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.0 GB · ready files0.5 GB
  • FP8 / INT8recommendedweights 0.0 GB0.5 GB
  • 4-bitweights 0.0 GB0.5 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.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — 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.

    # skyuu72/kibble-pythia-1b
    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 skyuu72/kibble-pythia-1b --local-dir ~/models/kibble-pythia-1b
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to kibble-pythia-1b
  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 pythia-1b (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

skyuu72/kibble-pythia-1b
Author
skyuu72
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments