Hugging Face
Open model and dataset hub.
Signals · 29
Open-sourcing AstaBrief, the fast report-generation model in Asta
Hugging Face has open-sourced AstaBrief, a fast report-generation model in the Asta ecosystem.
AutoSynthData: Generating Training Data for Enterprise Agents
Hugging Face announced AutoSynthData, a tool designed for generating synthetic training data for enterprise agents.
Introducing Olmo-core 3: Open, scalable training infrastructure for large MoEs
Hugging Face introduced Olmo-core 3, an open and scalable training infrastructure designed for large Mixture of Experts models.
NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction
Hugging Face highlighted NVIDIA Kumo Tabular setting a new accuracy and efficiency frontier in tabular prediction.
Getting the Source Right, Not Just the Fact: Source-Aware Verification for MCP Agents
Hugging Face discusses source-aware verification for MCP agents to ensure information sources are verified alongside facts.
Holo4: powering generalist computer-use agents
Hugging Face announced Holo4, a new model designed to power generalist computer-use agents.
Accelerating vision-language models with LFM2.5-VL-DSpark
Hugging Face announced a method to accelerate vision-language models using LFM2.5-VL-DSpark.
How to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning Workflows
A guide on using NVIDIA Warp and MjWarp to accelerate robotics simulation and learning workflows.
**Know Who Spoke When: Build Real-Time, Multi-Speaker AI with NVIDIA Nemotron 3 Diarization**
Hugging Face highlighted a guide for building real-time, multi-speaker AI systems utilizing NVIDIA's Nemotron 3 speaker diarization model.
Pruning LLMs Like a Physicist: Block Removal as an Ising Optimization Problem
Hugging Face presented a new approach to pruning large language models by framing block removal as an Ising optimization problem.
Your Agent Aced the Task. Will It Do It Again?
Hugging Face raises questions regarding the reliability and consistency of AI agents in repeating successful task execution.
IBM releases SOTA Granite Time Series PatchTST-FM-r2 model with commercial-friendly license
IBM has released its Granite Time Series PatchTST-FM-r2 model under a commercially friendly license.
Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic
Hugging Face discusses refining AI safety by refusing only harmful subsets of topics rather than blocking entire subjects.
NeoMME: an efficient Multimodal-native and Multilingual Encoder
Hugging Face introduced NeoMME, an efficient multimodal-native and multilingual encoder.
Real-Time Intelligence with IBM Time Series Models on Confluent
Hugging Face highlighted the integration of IBM time series models with Confluent for real-time intelligence.
BenchMIRT: What are LLM benchmarks actually measuring?
Hugging Face introduced BenchMIRT to investigate what large language model benchmarks are actually measuring.
Granite 4.2 LLMs: How They're Built
Hugging Face shares an overview detailing how the Granite 4.2 large language models are constructed.
Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original
Hugging Face announced a compressed 4-bit model that outperforms its full-precision original using quantization-aware healing.
Wire It, Run It, Deploy It: AI Workflows in Gradio
Hugging Face has introduced new capabilities for building, running, and deploying AI workflows within the Gradio framework.
Up to 3.2x Faster Inference with LFM2.5-DSpark
Hugging Face announced that LFM2.5-DSpark achieves up to 3.2x faster inference speeds.
LFM2.5 Q4\_0 Checkpoints from Quantization-Aware Distillation
Hugging Face released LFM2.5 Q4_0 checkpoints developed through quantization-aware distillation.
How Much Memory Does Your Agent Actually Need?
Hugging Face investigates the amount of memory required to run AI agents.
Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers
Hugging Face announced support for multi-vector late interaction embedding models using the Sentence Transformers library.
Same Cluster, 33 Points More Utilization: What Changed Was the Order
Hugging Face achieved a thirty-three point increase in cluster utilization by optimizing the execution order of workloads.
Record, train, and deploy from one place with Strands Agents, LeRobot, and Hugging Face Storage Buckets
Introducing OlmoEarth embeddings: Custom embedding exports from OlmoEarth Studio for downstream analysis