🌇Visual-language assistant using DeepSeek-VL2 and OpenVINO

🛠 DeepSeek-VL2 is an advanced series of large Mixture-of-Experts (MoE) Vision-Language models.

⚙️ DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.

🔰 More details can be found in the paper and original repository.

🔰 In this tutorial we consider how to convert and run DeepSeek-VL2 models using OpenVINO and optimize it using NNCF.

🔰 Table of contents:

🔹Prerequisites
🔹Convert model to OpenVINO 🔹Intermediate Representation
🔹Compress model weights to 4-bit
🔹Prepare inference pipeline
🔹Run model inference
🔹Visual Grounding
🔹Grounding Conversation
🔹Visual Question Answering
🔹Interactive demo

🔗Links:
https://github.com/openvinotoolkit/openvino_notebooks/blob/latest/notebooks%2Fdeepseek-vl2%2Fdeepseek-vl2.ipynb