Model Architecture: ViCLIP with ViT-L/14 vision tower. The vision and text encoders are both initialized from https://huggingface.co/laion/CLIP-ViT-L-14-DataComp.XL-s13B-b90K.

HF Transformers: The model is directly usable via AutoModel, AutoTokenizer and AutoVideoProcessor from HF Transformers.

Uses

This model is intended for research purposes. We hope it enables researchers to better understand and explore zero-shot video classification, video-text retrieval, and generalisation capabilities of video-language models.

Direct Use

Zero-shot video classification
Video-text retrieval
Video representation learning

Downstream Use

Linear probing and full fine-tuning for various video classification tasks
Video retrieval systems
Component for multimodal models (e.g., VLMs)

Out-of-Scope Use

Surveillance and facial recognition applications
Undeployed or unconstrained deployment without thorough in-domain testing
Any language other than English

Training Details

Training Data

Trained on the BVD-V-50M dataset.

Training Procedure

Model: ViCLIP L/14
Frames per video 8
Resolution: 224×224
Samples: 50M samples seen
Optimizer: AdamW (lr=4e-5, betas=[0.9, 0.98], weight_decay=0.2)
Global batch size: 32000
Warmup: 100 steps
Learning Rate Scheduler: Cosine annealing

Evaluation

Evaluated using the CLIP Benchmark video.

Results

Kinetics-400 top-1 accuracy	UCF-101 top-1 accuracy	HMDB51 top-1 accuracy	MSR-VTT video/text retrieval recall@1	MSVD video/text retrieval recall@1
63.3	79.7	61.4	42.8 / 43.2	53.7 / 83.9

How to Get Started with the Model

Quickstart with HF Transformers

python

import torch
import torch.nn.functional as F
from transformers import AutoModel, AutoTokenizer, AutoVideoProcessor

model = AutoModel.from_pretrained("laion/ViCLIP-L-14-BVD-V-50M-s50M-b32K", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("laion/ViCLIP-L-14-BVD-V-50M-s50M-b32K", trust_remote_code=True)
video_processor = AutoVideoProcessor.from_pretrained("laion/ViCLIP-L-14-BVD-V-50M-s50M-b32K", trust_remote_code=True)

video = video_processor("path/to/video.mp4")

labels = ["a person playing basketball", "a person swimming", "a person running"]
text_tokens = tokenizer(labels, padding="max_length", truncation=True, max_length=77, return_tensors="pt")

with torch.no_grad():
    v_feat = model.get_video_features(video["pixel_values_videos"])
    t_feat = model.get_text_features(text_tokens["input_ids"])

logits = 100.0 * v_feat @ t_feat.t()
probs = logits.softmax(dim=-1)
print("Label probs:", probs)

Citation

bibtex

@misc{laionbvd2026,
      title={LAION-BVD: A 10-Million-Hour Open Video Dataset for Multimodal Pre-training}, 
      author={Andreas Hochlehnert and Marianna Nezhurina and Mehdi Cherti and Andrej Radonjic and Thaddäus Wiedemer and Christoph Schuhmann and Romain Beaumont and Wieland Brendel and Bernhard Schölkopf and A. Sophia Koepke and Jenia Jitsev and Matthias Bethge},
      year={2026},
      eprint={2608.24845},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2608.24845}, 
}

laion/ViCLIP-L-14-BVD-V-50M-s50M-b32K

ViCLIP-L-14-BVD-V-50M-s50M-b32K

Table of Contents

Model Details