A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
Unknown
Molmo2 introduces open weights and data for vision-language models with video understanding and grounding, addressing efficient packing of crops and tokens.
Yue Fan, Xuehai He, Diji Yang, et al.
GRIT trains multimodal LLMs to generate reasoning chains interleaving text and bounding boxes, using RL with robust rewards to achieve data-efficient visual grounding.
Juan S. Santillana
VectraYX-Vision-1B is a sub-2B Spanish/LATAM cybersecurity VLM with structured reasoning and tool use, but current vision SFT yields near-zero visual grounding.
Unknown
This paper introduces grounded chain-of-thought (GCoT), a new learning task that enhances multimodal large language models by grounding their reasoning steps in visual evidence.
Ayoub Kirouane, Christos Petrocheilos
End-to-end adaptation of NVIDIA's Nemotron retrieval stack for Modern Greek, including a new benchmark HERA, achieving large gains in retrieval and grounded generation.
Zhiliang Peng, Wenhui Wang, Li Dong, et al.
Kosmos-2 introduces grounding capability to multimodal large language models, enabling them to perceive object descriptions and ground text to the visual world.
Boyang Zhang, Adrian Lyjak, Eli Stewart, et al.
ExtractBench is a benchmark for schema-guided enterprise document extraction, evaluating value accuracy, record completeness, grounding, and cost across 370 documents.
Bing Yan, Gregory Wolfe, Stefano Martiniani, et al.
AskChem is a claim-centered infrastructure for chemistry literature search that converts papers into atomic, provenance-carrying claims, enabling cross-paper synthesis and grounding AI agents.
Unknown
DataGemma reduces LLM hallucinations by grounding responses in Google's Data Commons statistical database using RIG and RAG methods.
Unknown
LMDX adapts arbitrary LLMs for document information extraction with layout encoding and grounding to prevent hallucinations.
Unknown
This paper unlocks referring and grounding capabilities in multimodal large language models to create a more flexible human-computer interface.
Unknown
IALP system integrates grounding mechanisms into long-horizon planning for embodied mobile manipulation.