furiosa-ai/Qwen3-VL-2B-Instruct
Qwen3-VL-2B-Instruct by Qwen is a 2-billion-parameter dense vision-language model from the Qwen3-VL series, designed for multimodal tasks. It integrates a vision encoder with a dense transformer decoder, utilizing Interleaved-MRoPE positional embeddings and DeepStack multi-level feature fusion. This Instruct edition excels at visual understanding, including OCR, document/chart analysis, spatial reasoning, video comprehension, and natively supports tool calling, making it suitable for complex multimodal AI applications.
Loading preview...
Overview
Qwen3-VL-2B-Instruct is a 2-billion-parameter dense vision-language model from the Qwen3-VL series, developed by Qwen. It combines a vision encoder with a dense transformer decoder, featuring Interleaved-MRoPE positional embeddings and DeepStack multi-level feature fusion. This model is the "Instruct" edition, designed for direct instruction following in multimodal contexts.
Key Capabilities
- Vision-language understanding: Processes both images and text, supporting tasks like OCR, document analysis, chart interpretation, spatial reasoning, and video comprehension.
- Tool calling: Natively supports tool (function) calling via the
hermestool-call parser, enabling interaction with external functions. - Multimodal chat: Accepts OpenAI-style multimodal chat messages, including
image_urlcontent parts alongside text. - Hardware optimization: This specific repository provides a Furiosa Executable Bundle (FXB) for efficient deployment on FuriosaAI RNGD hardware using Furiosa-LLM, running with a tensor-parallel size of 8 PEs on a single RNGD card.
Good for
- Applications requiring robust visual understanding combined with natural language processing.
- Developing AI agents that can interact with tools based on multimodal inputs.
- Deploying multimodal models on FuriosaAI RNGD hardware for optimized inference performance.
- Building interactive chat systems that can interpret and respond to image-based queries.