Skip to content
Other

prelude-base 100%

YaRN 160k: окно 160K без дообучения

1 min readupdated

Это финальный чекпоинт openeurollm/prelude-base (ревизия anneal2t_multilingual1_dominant1_iter_1192093): 100% претрейна, около 10T токенов, 9,09 млрд параметров. Архитектура Qwen3-style dense: 36 слоёв, hidden 4096, FFN 12288, 32 головы внимания / 8 KV-групп. Веса не менялись — только config.json с YaRN RoPE для окна 163840 токенов; обучение не проводилось.

Ключевые особенности

  • YaRN: factor 11.9, rope_theta 813000, max_position_embeddings 163840.
  • Полезное окно — 163840 токенов; прирост NLL на коротком контексте +0.31.
  • Perplexity: 4.85 в полосе 64K–128K и 6.45 в 128K–160K.
  • За 163840 уходить нельзя: следующая полоса сломана, ppl 235.
  • Конфиг читают и старые/новые transformers, и vLLM.
  • Это base-модель без instruction tuning и safety alignment.

Пример

vllm serve <model> --max-model-len 163840
ПодробнееЗаменяет ранний 128K-конфиг: тот же короткий контекст, но окно ломается сразу после 131072.
Требования к VRAM в карточке не указаны.
Это только репараметризация RoPE, без continued pretraining; downstream-бенчмарки и retrieval-тесты не проводились.
Если дообучаете, сохраняйте этот конфиг: веса адаптируются к параметризации.
Discuss Telegram 1 17

Will it run?

Working out the memory needs

The bot is still collecting this model's weight sizes; the estimate will appear here in a few minutes. Meanwhile you can rent a GPU with headroom.

Massed ComputeOpen Massed Compute

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.

    # ankush13r/prelude-base-iter_1192093-yarn-160k
    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 ankush13r/prelude-base-iter_1192093-yarn-160k --local-dir ~/models/prelude-base-iter_1192093-yarn-160k
    # The model is in ~/models. How to run it is in the model card (link above).
  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 prelude-base-iter_1192093-yarn-160k (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

ankush13r/prelude-base-iter_1192093-yarn-160k
Author
ankush13r
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

171

Ask AI about this model

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

Comments