Vit Pe Lang Large Patch14 448.fb

--- tags: - image-feature-extraction - timm - transformers library_name: timm license: apache-2.0 --- # Model Details This is a `timm` remapped, image encoder only variant of the original weights. [\[📃 Tech Report\]](https://arxiv.org/abs/2504.13181) [\[📂 Github\]](https://github.com/facebookresearch/perception_models/) Perception Encoder (PE) is a state-of-the-art encoder for image and video understanding trained via simple vision-language learning. It was introduced in "[Perception Encoder: The best visual embeddings are not at the output of the network](https://ai.meta.com/research/publications/perception-encoder-the-best-visual-embeddings-are-not-at-the-output-of-the-network/)". **Model Developer**: Meta **Model Overview**: Perception Encoder (PE) is a family of large-scale vision encoder models with state-of-the-art performance on a large variety of vision tasks. By using a robust contrastive pretraining recipe and finetuning on synthetically aligned videos, PE not only outperforms all existing models on classification and retrieval, but it also internally produces strong, general features that scale for downstream tasks. PE unlocks the ability for large-scale contrastive pretraining to transfer to downstream tasks with alignment tuning to capitalize on those general features. <img src="https://huggingface.co/facebook/PE-Core-G14-448/resolve/main/docs/pe_image1.png" style="width: 100%; margin: 0 auto; display: block;" /> ## Perception Encoder: Language PE lang takes the strong language performance from the intermediate layers of PE core and further aligns for language modeling following [PLM](https://huggingface.co/papers/2504.13180). We specifically tuned PE lang to be versatile for any multimodal langugage modeling use case, including using different language model decoders (e.g., Llama / Qwen) and using different eval settings (e.g., native res / tiling). PE lang performs particularly well on OCR and document tasks. We release two PE Lang checkpoints, L14-448 and G14-448. Here are their results our benchmark setting with frozen encoder with 2.6M SFT datamix, using 448px _only_ (i.e., _with no tiling_) and Llama 3.1 8B as the decoder: | Encoder | Checkpoint | Doc VQA (val) | InfoQA (val) | TextVQA | MVBench | PerceptionTest (val) | EgoSchema (val) | |:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:| | **L/14** 448px | [PE-Lang-L14-448](ht...