LLMs & Generative AI - Page 11 of 55
Latest breakthroughs in large language models and generative AI shaping the future of artificial intelligence and machine learning.
1086 articles
View complete article list
Errorquake-10k Benchmark Scores 10,000 LLM Responses on 0-4 Severity Scale
Accuracy alone is a lie. It flattens every mistake into a single number, hiding the gulf between a model misremembering a date and one concocting a false patient history. The Errorquake-10k benchmark shatters that illusion.
Three SpaCy Tricks Speed Up Production-Grade Text Processing
Processing text at scale doesn't have to be slow. Most spaCy pipelines handle documents individually, a method that wastes CPU cycles and complicates data alignment.
Zhipu AI employs Muon Optimizer and Muon Split in GLM-4.5 and GLM-5 pretraining
The wall isn't made of silicon. At Zhipu AI, engineers hit it while training GLM-4.5. The problem was the optimizer—the software that tweaks a model's billions of internal knobs. Adam, the industry standard, had stalled.
Anthropic says Claude writes >90% of its code; AI pause button urged
Anthropic dropped a figure on Wednesday that stops you cold. Its AI assistant Claude now authors more than 90 percent of the company's own production code. That's not a trial run—it's the core of how a $18 billion firm operates.
Choosing AI Models: Prioritize Real‑World Needs Over Benchmark Rankings
Everyone's talking about AI model leaderboards. Almost no one should care. Your project doesn't need the best model in the world. It needs the one that works. Benchmarks are built on synthetic tests.
ELI releases LLM benchmark showing top models resist Russian propaganda
Estonia knows the weight of a neighbor’s lie. A former Soviet republic, it has spent decades untangling narratives spun from Moscow.
AI trust certification trial in Fintech, Banking, Insurance, Health, US, Vietnam
Putting an AI agent in charge of a loan application or a patient triage system requires a leap of faith few executives are ready to make. A new pilot program tried to replace that faith with something better: hard numbers from a simulated gauntlet.
SMAC-Talk Adds Natural Language to StarCraft Multi-Agent Challenge for LLMs
The Starcraft Multi-Agent Challenge just got a backstabbing social layer. A new benchmark called SMAC-Talk forces AI agents to coordinate using natural language chat. The twist is that one agent might be lying.
Spectral transfer identity s=αγ ties curvature exponent to Hessian decay
The curvature of a neural network’s loss landscape is not a monolith, it decomposes.
ChatHealthAI Aligns Structured EHR Data with Frozen LLM for Clinical Reasoning
Doctors relying on AI for predictions often face a black box: a risk score appears, but the reasoning behind it stays hidden. A research team has now built ChatHealthAI to tackle that opacity.
Study Explores Graph Scaffolds as Reasoning Aid for Large Language Models
AI research is drowning in text. A new study suggests the solution might be lines and boxes. The idea is simple. Humans don't just think in paragraphs. When a problem gets complex, we sketch. We make mind maps, flowcharts, diagrams.
NVIDIA releases Cosmos 3 with Super‑Text2Image and Nano‑Policy‑DROID
NVIDIA’s Cosmos 3 isn’t just another model drop. It’s a two-tower mixture-of-transformers foundation model that fuses physical reasoning, world generation, and action generation into a single unified framework.
Guide: Run a Claude Managed Agent Task End‑to‑End via Session Stream
The moment you connect a model to a runtime sandbox, you unlock something far more powerful than a chatbot answering questions.
Microsoft unveils Surface NVIDIA RTX Spark Dev Box for AI agent development
Microsoft’s latest play isn’t just a new box. It’s a hardened launchpad for AI agents that work where you work: on your desk, in your local environment.
Nvidia builds RTX Spark supercomputer chips with Microsoft for AI agents
Every hardware company is now an AI company. Nvidia, with help from Microsoft, just built the chip to prove it. The RTX Spark is a supercomputer part designed to run AI agents on a regular PC. This is not a small step.
LLM-derived valence direction aligns with EEG signals in 123 subjects
The simplest measure of emotion, valence, the axis from pleasure to displeasure, has long resisted a clean neural readout.
gSMILE Framework Tackles LLM Transparency by Mapping Prompt Responses
Large language models are famously opaque. Their reasoning happens somewhere between the question you type and the answer you get, a process that's hidden and unmarked. The gSMILE framework wants to change that.
DAStatFormer extracts 24 ANOVA-selected features per channel, slashing data size
Every new sensor dumps a tidal wave of data onto the shore. Most of it is useless static. The real breakthrough, argues a team on arXiv, isn't in building a bigger server to process the flood, but in installing a smarter filter at the source.
Test-Time Prompt Optimization Turns Demonstrations into Rewards for VLM Models
Getting a reinforcement learning agent to work often comes down to one frustrating task: engineering its reward signal. Now, researchers are pressing general-purpose Vision-Language Models into service as reward judges.
BitsMoE uses SVD to keep basis unquantized, allocating bits to expert spectral factors
The arithmetic of mixture-of-experts models promises efficiency, but quantization, the brutal necessity of shrinking them for deployment, has always demanded a cruel trade-off: compress the shared structure or starve the expert nuances.
Featured Resources & Reviews
No Code MBA Course Review
Learn to build AI-powered apps without coding. Our comprehensive review of No Code MBA's course.
AI Tools & Resources
Curated collection of AI tools, courses, and frameworks to accelerate your AI journey.
Weekly AI Digest
Get the week's most important AI news delivered to your inbox every week.