Skip to content
Other

LeVJEPA 1.5 Large

видеоэнкодер без лишних эвристик

2 min readupdated

LeVJEPA 1.5 Large от galilai-group — self-supervised видеоэнкодер на 303M параметров (ViT-L/16). Его обучали на ~5 300 часах видео и 151 EF, а средний балл на 22 image/video датасетах составил 78.76%.

Ключевые особенности

  • 303M параметров, ViT-L/16: 24 слоя, ширина 1024, 16 голов; на вход 16 кадров 224×224.
  • Один энкодер с invariance loss по глобальному и локальным view клипа + SIGReg; без EMA target encoder, stop-gradient, predictor и pixel decoder.
  • 78.76% на 22 датасетах при ~5 300 часах видео и 151 EF; V-JEPA 1 — 79.77% при 25.7× данных и 5.6× compute, V-JEPA 2 — 81.63% при 321× данных и ~28.5× compute.
  • Замороженный энкодер для attentive/linear probing, retrieval или backbone; classification head нет.
  • Block-causal attention: bidirectional внутри кадра, causal между кадрами; [cls] читает весь клип.

Пример

import torch
from transformers import AutoModel

model = AutoModel.from_pretrained('galilai-group/LeVJEPA-1.5-Large', trust_remote_code=True).eval()
video = torch.randn(1, 3, 16, 224, 224)  # (B, C, T, H, W)
with torch.no_grad():
    out = model(pixel_values=video)
ПодробнееСредний балл: 78.76 на 22 датасетах; image mean 83.34, video mean 68.94.
Обучение: ~5 300 часов видео (1.8M клипов), 151 EF; 95% patch-токенов случайно отбрасываются; веса — EMA-копия.
Ограничения: нет classification head; лицензия CC BY-NC 4.0 (некоммерческая); full attention даёт худшие признаки.
Требования к VRAM/железу в карточке не указаны.
Discuss Telegram 15 27

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.

    # galilai-group/LeVJEPA-1.5-Large
    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 galilai-group/LeVJEPA-1.5-Large --local-dir ~/models/LeVJEPA-1.5-Large
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to LeVJEPA-1.5-Large
  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 LeVJEPA-1.5-Large (quants, finetunes, LoRAs) will show up under the bell.

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

About the model

galilai-group/LeVJEPA-1.5-Large
Author
galilai-group
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

2715

Ask AI about this model

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

Comments