AI Briefing#vlm
The latest AI and developer news about #vlm, with the original source and a short summary.
The latest AI and developer news about #vlm, with the original source and a short summary.
This guide explains how to optimize and run VLMs on Intel CPUs using OpenVINO and Optimum Intel.
A two-stage training methodology and dataset for evolving lightweight VLMs into GUI automation agents has been released.
This covers a method for maximizing the analysis performance of Vision Language Models (VLM) through fine-tuning that reflects the unique characteristics of satellite imagery.
HuggingFace's TRL library now supports the latest alignment techniques such as MPO and GRPO for VLMs.
NVIDIA has released Llama Nemotron Nano, an 8B-scale VLM with enhanced document understanding and OCR performance.
ScreenSuite, an integrated benchmark suite for evaluating GUI agent performance from multiple angles, has been launched.
Holo1, an open-source VLM family specialized for GUI automation, and the web agent Surfer-H have been released.
Hugging Face has released nanoVLM, a lightweight toolkit that lets even beginners easily train Vision Language Models.
This summary covers the major technology trends, latest models, and benchmarks in Vision Language Models (VLM) that have rapidly evolved over the past year.
Intel has released AutoRound, a tool for LLMs/VLMs that maintains high accuracy even at low-bit quantization.
Shares how olmOCR was fine-tuned to extract business document headers and footers without omission, along with the results.
Cohere For AI has released the 'Aya Vision' series, an open-weight multimodal model supporting 23 languages.
Google has launched the PaliGemma 2 Mix models, optimized for a variety of vision-language tasks.
Qwen has released Qwen2.5-VL, its next-generation vision-language model with agentic capabilities and long-video understanding.
HuggingFace's smolagents has been updated with VLM support, enabling it to process visual information.
HuggingFace has released SmolVLM 256M and 500M, ultra-lightweight multimodal models.
HuggingFace has released SmolVLM, a 2B-scale open-source vision language model that maximizes memory efficiency.
It introduces a multimodal data augmentation technique that simultaneously transforms text and images in document images to improve VLM training efficiency.
HuggingFace's TRL library now supports Direct Preference Optimization (DPO) for Vision Language Models (VLMs).
TII has released the 11B-scale Falcon 2 LLM and VLM, trained on over 5 trillion tokens.
Google has released PaliGemma, a new open vision language model (VLM) that combines SigLIP and Gemma.
Hugging Face has released Idefics2, an 8B-scale open source multimodal model with strong OCR performance.
A guide summarizing the principles and key models of Vision Language Models (VLMs), which process images and text simultaneously.
A large-scale synthetic dataset called WebSight and a model called Sightseer, which convert screenshots into HTML code, have been released.