Numind has released NuExtract3, a 4B open-weight VLM based on Qwen3.5-4B under an Apache-2.0 license. It aims to make information extraction from complex documents like PDFs, screenshots, and forms more practical and accessible.
![NuExtract3 released: open-weight 4B VLM for Markdown, OCR and structured extraction (self-hostable) [P]](/cdn-cgi/image/width=3840,quality=75,format=webp/https://preview.redd.it/pm2xbooyxn2h1.png?width=140&height=78&auto=webp&s=30ec102187e4eb4582c79f124c41b8a3c129d1fd)
The content asks for opinions on the suitability of EMNLP workshops for Vision-Language Model (VLM) evaluation work. It also seeks recommendations for other good NLP venues for this type of research.
This paper introduces MedExpMem, an experience memory framework designed to enhance medical vision-language models (VLMs) with differential diagnosis expertise. It allows diagnostic agents to learn from their own failures by memorizing discriminative experiences as pairwise differential notes.
This guide addresses the repetitive retraining of object detection models like YOLO in industrial settings by proposing Generative Vision-Language Models (VLMs) for zero-shot detection. It highlights how VLMs transform detection into semantic prompting, bypassing continuous data collection and retraining, but notes new architectural challenges for industrial engineering teams.
PRISM is a novel framework that integrates perception (VLM) and decision (LLM) through a dynamic question-answer pipeline, enabling the LLM to actively refine the VLM's output for task-driven scene understanding. This approach significantly outperforms existing image-based models on benchmarks like ALFWorld and Room-to-Room.
Pelican-Unified 1.0 is introduced as the first embodied foundation model, integrating a single VLM for understanding, imagination, and action generation in one forward pass. This unified approach outperforms modular systems on benchmarks and simplifies robotics development.