Skip to content
Other

PrimeSim

мультивью world model для симуляции робота

2 min readupdated

PrimeSim — action-conditioned multiview world model, который работает как симулятор робота. На вход получает первый кадр с каждой камеры, промпт и траекторию действий, а на выходе генерирует продолжение. Модель от qiyuan-robot на базе Wan-AI/Wan2.2-TI2V-5B: 5B transformer, библиотека diffusers, обучение на 30 110 часах cross-embodiment данных с 22 платформ.

Ключевые особенности

  • Генерирует три камеры одновременно в 192×256 и работает выше 10 FPS.
  • 5B transformer на базе Wan2.2-TI2V-5B; есть код inference и training.
  • Обучен на 30 110 часах данных с 22 платформ; в симуляции лидирует в 4 из 5 suite WorldSimProbe и показывает лучший общий EWMBench.
  • Два чекпоинта: primesim_base.safetensors для fine-tune и primesim_posttrain_g2.safetensors (causal few-step) для inference.
  • Веса не self-contained: text encoder, VAE и tokenizer нужно скачать из Wan2.2-TI2V-5B и прописать пути в config/model/primesim.yaml.
  • Apache 2.0: веса открыты, но training corpus не распространяется.

Пример

Сначала подтяните недостающие компоненты из базовой модели:

hf download Wan-AI/Wan2.2-TI2V-5B \
  --include "models_t5_umt5-xxl-enc-bf16.pth" "Wan2.2_VAE.pth" "google/umt5-xxl/*" \
  --local-dir checkpoints

Затем укажите три полученных пути в config/model/primesim.yaml.

ПодробнееБенчмарки: 4 из 5 suite WorldSimProbe и лучший общий EWMBench.
Железо: рекомендуется A100 или H100 80 GB — 5B transformer и 11 GB text encoder одновременно не влезают в 24 GB.
Ограничения: веса не self-contained, training corpus не распространяется.
Лицензия весов: Apache 2.0.
Discuss Telegram 1 120

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.

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

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

About the model

qiyuan-robot/PrimeSim
Author
qiyuan-robot
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

1201

Ask AI about this model

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

Comments