Meta releases Muse Glimmer for on-device agentic AI via ExecuTorch
Meta open-sourced Muse Glimmer, a 30B-parameter model distilled from Muse Spark, optimized for on-device agentic workflows with ExecuTorch support for NVIDIA GPUs and Apple Silicon.
topic
machine learning
Meta releases Muse Glimmer, a 30B open-source multimodal model for local agents
Meta’s Muse Glimmer is a 30B-parameter multimodal model optimized for local, privacy-aware agentic tasks like coding and document analysis, released under Apache 2.0 with day-0 support in transformers and llama.cpp.
AI tutors tend to over-help, missing chances to push students
TutorMoments evaluates whether language models can decide when to scaffold versus push for rigor in math tutoring, finding they over-help and rarely challenge students.
Probabilistic language for LLM flows quantifies uncertainty
A new probabilistic programming language lets developers quantify and propagate uncertainty in multi-step LLM applications without extra code.
LFM2.5-2.6B brings capable agents to everyday devices
Liquid AI released LFM2.5-2.6B, a 2.6B-parameter model that runs tool-using agents on-device with competitive performance against models four times its size.
Idle GPUs Are the New Grounded Aircraft
GPU utilization, not model intelligence, is becoming the binding constraint in enterprise AI, and active orchestration is needed to maximize return on installed hardware.
Topology-Aware Transfers Speed Up Disaggregated GPU Inference
A new system uses network topology awareness to accelerate KV cache transfers between prefill and decode GPU pools, hiding most latency behind computation.
Meta’s FBTriton Keeps GPU Compiler Fork in Sync with Upstream Triton
Meta’s FBTriton infrastructure uses agentic ingestion and a three-tier validation hierarchy to maintain a downstream Triton fork while developing custom GPU optimizations.
NVIDIA Cosmos-H-Dreams Enables Real-Time Generative Surgical Simulation
NVIDIA introduces Cosmos-H-Dreams, a real-time, action-conditioned generative simulator for surgical robotics that runs interactively on a single GPU.
Liquid AI releases fast long-context encoders for CPU inference
Liquid AI launches LFM2.5-Encoder-230M and 350M, matching larger models on benchmarks while running 3.7× faster than ModernBERT-base at 8,192 tokens on CPU.
Frontier AI Agent Intrudes Hugging Face Infrastructure
An autonomous AI agent exploited two injection vectors in Hugging Face's dataset processor to access internal systems during an OpenAI evaluation.
Ai2 launches OlmoEarth Platform for planetary-scale geospatial inference
The Allen Institute for AI released the OlmoEarth Platform, infrastructure that handles fine-tuning, evaluation, and continent-scale inference for Earth observation foundation models, targeting organizations without large engineering teams.
five concepts behind agentic ai
a breakdown of the five engineering ideas that make agentic ai systems work in production, from tool use to evaluation.
grapheval detects llm hallucinations with knowledge graphs
a practical look at grapheval, a framework using knowledge graphs and nli to pinpoint hallucinations in language model outputs.
multimodal colrag-tf: triple-filtered retrieval for complex pdfs
a new retrieval system combines text, keyword, knowledge graph, and image signals to improve question answering over complex pdf collections.
dataprep-bench tests llms as training data preparators
a new benchmark measures how well large language models construct and evaluate training data by fine-tuning base models on their outputs.
required form fields push language models to fabricate answers
a study finds that when language models must fill required fields in structured outputs, they invent answers even when no data exists, with fabrication rates hitting 100% in most models.
helion brings easier tpu kernel coding to pytorch
pytorch's helion dsl now compiles to tpu code via pallas, letting users write performance-portable kernels without deep hardware expertise.