Skip to main content
← All Tags

Machine Learning

273 articles in this category (Page 3 of 12)

AI NewsMachine LearningData Science

TabPFN vs. CatBoost: Achieving Superior Tabular Accuracy with In-Context Learning

TabPFN achieves 98.8% accuracy on tabular datasets using in-context learning, outperforming CatBoost and Random Forest with near-zero training time.

Read more
AI NewsMachine LearningOpen Source

OpenMythos: A 770M Parameter Recurrent-Depth Transformer Matching 1.3B Models

OpenMythos reconstructs Claude Mythos using Recurrent-Depth Transformer architecture, enabling a 770M parameter model to match 1.3B parameter performance.

Read more
AI NewsLanguage ModelMachine Learning

PrfaaS: Scaling LLM Serving via Cross-Datacenter Prefill-as-a-Service

Moonshot AI and Tsinghua's PrfaaS architecture boosts LLM serving throughput by 54% using cross-datacenter KVCache transfer over commodity Ethernet.

Read more
AI NewsMachine LearningSoftware Engineering

Deep Dive into Transformer Architectures: Stacking Self-Attention Layers for Context

Transformer models transition from positional encodings to stacked self-attention layers to capture deep contextual relationships in complex text.

Read more
AI NewsArtificial IntelligenceMachine Learning

Subliminal Learning: How LLMs Inherit Hidden Behavioral Traits via Synthetic Data

New research in Nature reveals student LLMs inherit teacher behavioral traits through hidden signals in synthetic data, even when datasets are semantically unrelated.

Read more
AI NewsArtificial IntelligenceMachine Learning

NVIDIA and University of Maryland Release Audio Flamingo Next (AF-Next)

NVIDIA's AF-Next outperforms Gemini 2.5 Pro on LongAudioBench with a 73.9 score, scaling open audio reasoning to 1 million hours of data.

Read more
AI NewsArtificial IntelligenceMachine Learning

Building Privacy-First AI Agents with Gemma 4 and Ollama

Build a local tool-calling agent using Google’s Gemma 4:e2b model and Ollama to execute Python functions with zero latency and high privacy.

Read more
AI NewsLanguage ModelsMachine Learning

Structured Outputs vs. Function Calling: Architectural Trade-offs for AI Agents

Learn the architectural differences between structured outputs and function calling to build reliable AI agents with 100% schema compliance.

Read more
AI NewsArtificial IntelligenceMachine Learning

Meta AI and KAUST Propose Neural Computers: Folding Computation and Memory into One Learned Model

Meta AI and KAUST researchers introduce Neural Computers (NCs), achieving 98.7% cursor accuracy in GUI prototypes by folding OS functions into a single learned runtime state.

Read more
AI NewsDeep LearningMachine Learning

Knowledge Distillation: Compressing Ensemble Intelligence for Efficient AI Deployment

Learn how knowledge distillation recovers 53.8% of an ensemble's accuracy edge while achieving 160x model compression for production.

Read more
AI NewsArtificial IntelligenceMachine Learning

Sigmoid vs ReLU: Why Geometric Context Preservation is Critical for Neural Network Inference

ReLU outperforms Sigmoid by preserving geometric distance from decision boundaries, achieving 96% accuracy compared to Sigmoid's 79% in two-moons benchmarks.

Read more
AI NewsAI InfrastructureMachine Learning

Five AI Compute Architectures Every Engineer Should Know: CPUs, GPUs, TPUs, NPUs, and LPUs Compared

Understand the trade-offs between AI architectures, including Groq’s LPU which achieves 10x higher energy efficiency than traditional systems for LLM inference.

Read more
AI NewsAgentic AIMachine Learning

Google AI Research Introduces PaperOrchestra: A Multi-Agent Framework for Automated AI Research Paper Writing

Google AI Research debuts PaperOrchestra, a multi-agent system that transforms raw experimental logs into submission-ready LaTeX papers, achieving simulated acceptance rates of up to 84%.

Read more
AI NewsMachine LearningAI Research

Extracting Emergent Structural Knowledge from LLMs through Sideways Questioning

Sean Trifero explores Eliciting Latent Knowledge (ELK) to uncover cross-domain structural patterns encoded within billions of LLM parameters.

Read more
AI NewsComputer VisionMachine Learning

Meta AI's EUPE: A <100M Parameter Universal Vision Encoder Rivaling Specialists

Meta AI introduces EUPE, a compact vision encoder under 100M parameters that matches domain-expert models in classification and dense prediction, achieving 55.2ms latency on iPhone 15 Pro.

Read more
AI NewsArtificial IntelligenceMachine Learning

MaxToki: A 1B-Parameter Temporal Foundation Model for Cellular Aging Trajectories

MaxToki, a 1B-parameter transformer, predicts cellular aging trajectories by training on 1 trillion gene tokens to identify disease-related age acceleration.

Read more
AI NewsMachine LearningSoftware Engineering

Engineering Production-Ready RAG Pipelines: Lessons from the Python Ecosystem

Learn how to move RAG from prototype to production using Python, FAISS, and SentenceTransformers while managing latency and data consistency for datasets under 100,000 chunks.

Read more
AI NewsAIMachine Learning

Decoding Attention Mechanisms: Final Steps and the Shift to Transformers

Learn how unrolling LSTMs and applying softmax similarity scores allows models to reach the EOS token in the final stage of decoding.

Read more
AI NewsAgentic AIMachine Learning

Google DeepMind AlphaEvolve: LLM-Driven Evolutionary Search Outperforms Human-Designed Game Theory Algorithms

DeepMind's AlphaEvolve uses Gemini 2.5 Pro to evolve MARL source code, discovering algorithms that outperform expert-designed baselines in 10 of 11 test games.

Read more
AI NewsAI InfrastructureMachine Learning

Optimizing Deep Learning Models with NVIDIA Model Optimizer and FastNAS Pruning

Learn how to build an end-to-end optimization pipeline using NVIDIA Model Optimizer and FastNAS to reduce ResNet20 complexity to a 60M FLOPs target.

Read more
AI NewsMachine LearningSaaS

Optimizing AI Sales Agents with Real-Time Intent Data and MCP Servers

Boost AI SDR response rates from 3% to 25% by integrating live intent data APIs to eliminate the 30% annual decay of static contact databases.

Read more
AI NewsAI InfrastructureMachine Learning

Hugging Face Releases TRL v1.0: A Unified Post-Training Stack for SFT, Reward Modeling, DPO, and GRPO Workflows

Hugging Face TRL v1.0 standardizes LLM post-training with a unified CLI and config system, delivering up to 2x training speed and a 70% reduction in memory usage.

Read more
AI NewsMachine LearningSoftware Engineering

Mastering Mixture of Experts: Scaling Large Language Models via Sparse Architectures

The Mixture of Experts (MoE) paradigm reduces inference compute costs by activating specialized sub-networks instead of monolithic dense parameters.

Read more
AI NewsAgentic AIMachine Learning

Z.ai GLM-5V-Turbo: Native Multimodal Vision Model for Agentic Engineering

Zhipu AI (Z.ai) launches GLM-5V-Turbo, a native multimodal vision coding model featuring a 200K context window and optimized integration for OpenClaw agentic workflows.

Read more