# Cross-modal Understanding

InternVL3 is an advanced multimodal large language model series with powerful multimodal perception and reasoning capabilities, supporting image, video, and text inputs.

Cephalo-Gemma-3-4b is a vision-language model specialized in biomaterials and spider silk analysis, fine-tuned based on the Gemma architecture.

Qwen2.5-Omni is an end-to-end multimodal model capable of perceiving various modalities such as text, images, audio, and video, and generating text and natural speech responses in a streaming manner.

Centurio is an open-source multilingual large vision-language model supporting 100 languages, capable of processing image-to-text and text-to-text tasks.

An encoder-decoder model based on CLIP encoder and GPT2 architecture for generating Thai image descriptions

Meta Chameleon is a hybrid-modal early fusion foundation model developed by FAIR, supporting multimodal processing of images and text.

This model is an image-to-text model based on the Apache-2.0 license, capable of converting image content into textual descriptions.

BLIP is a unified vision-language pretraining framework, excelling in image caption generation and understanding tasks, efficiently utilizing web data through guided annotation strategies

This is an image-to-text model based on the Apache-2.0 license, capable of converting image content into textual descriptions.

This is a CLIP ViT-B/16 model trained using OpenCLIP on the DataComp-1B dataset, primarily used for zero-shot image classification and image-text retrieval.

Pix2Struct is an image encoder-text decoder model trained on image-text pairs, supporting various tasks including image captioning and visual question answering.

This is an image-to-text model based on the MIT license, capable of converting image content into textual descriptions.

Microsoft's VinVL foundational pre-trained model, specifically designed for image captioning tasks, with strong visual-language understanding capabilities.

Chinese CLIP is a simplified implementation of CLIP based on approximately 200 million Chinese image-text pairs, using ViT-L/14@336px as the image encoder and RoBERTa-wwm-base as the text encoder.

molt5-base is a model based on the T5 architecture, specifically designed for translation tasks between molecules and natural language.