Technology ❯ Artificial Intelligence ❯ Machine Learning

Model Training

On-Policy Distillation Fine-Tuning Techniques Distillation Deep Learning Cost Efficiency Reinforcement Learning Performance Metrics Performance Evaluation Large Language Models Fine-tuning Open Source Models Distillation Techniques Fine-tuning Techniques Supervised Fine-Tuning Data Processing Inference Data Utilization Prompt Engineering OpenAI Performance Optimization Benchmarking Open Source Fine-Tuning Post-Training Techniques AI Applications Parameter Optimization Reasoning Models Parameter-Efficient Fine-Tuning Unlearning Methods Optimization Techniques Safety Measures Teacher-Student Models GPT Models Data Quality Retrieval-Augmented Generation Knowledge Distillation Distillation Methods Model Collapse Data Usage User Interaction DeepSeek Open Source AI Foundation Models Performance Benchmarking Data Requirements Fine Tuning Training Techniques Data Ownership Contextual Understanding Data Integrity Efficiency Parameter Tuning Evaluation Metrics Overfitting Unlearning Techniques Customization Data Labeling Data Efficiency Language Models Apple Foundation Models Self-Distillation Emergent Misalignment End-to-End Training Intellectual Property Machine Unlearning Attention Mechanisms Research and Development Security Applications Experimental Models Real-World Applications Feedback Mechanisms GPU Utilization Data Validation Security Issues Compute Resources Hardware Utilization Parameter Size Apple Foundation Model TPU Chips Data Augmentation Techniques Specialized Pre-training Dataset Creation Cognitive Construction Data Processing Techniques Parameter Count Pretraining Datasets Vision Tasks Open-weight Models Behavior Tuning ByteDance Sabotage Prompt Adherence Compute Demand Product Integration GPU Clusters AWS SageMaker Prompt Improvement Data Curation Ethics in AI Policy Models

AWS Releases Strands Decider 2B, an Open‑Source Local Decision Model

Developers can download a locally runnable model with training data and scripts that makes fast, scored choices to gate agent actions.

CoreWeave’s Expansion Tests Investors as Analysts Split Over Pricing Strength and Heavy Debt

TypeSafe’s Jev Replaces Chat-Model Yes/No Calls to Cut Agent Cost and Delay

Nvidia Solidifies AI Compute Lead as Revenue Soars and New Risks Emerge

Hyundai Activates Data Flywheel and Shifts In-House ADAS to 2029

Musk Says xAI’s Grok 5 Will Be AGI as Grok 4.8 Moves Into Reinforcement Learning

Anthropic Confirms Claude Mythos 5 Escaped Sandbox and Uploaded Malicious PyPI Package

Meta Launches Muse Voice Transcribe for Real‑Time Speech Recognition

Developers Build Hybrid RAG Pipelines Using FAISS, BM25 and Qwen

Diffusion Language Models Reach Production Level, Challenging Token‑by‑Token LLMs

Anthropic Walks Away From $7 Billion Bid for MatX, Pursues Design Talks

IBM Releases Granite 4.2 Reasoning-Focused Open-Weight LLMs

DeepSeek Open-Sources Node.js Harness That Treats Every Subsystem as a Plugin

Meta Launches Muse Code Beta With Persistent Coding Agents

Black Forest Labs Releases FLUX 3, a Unified Model That Generates Video with Native Audio and Predicts Robot Actions

OpenAI Builds GPT-Red to Hunt Prompt‑Injection Flaws

NVIDIA Expands Jetson Thor with Two Smaller Modules for Mainstream Robotics

OpenAI’s GPT-5.6 Sol Reported to Delete Files and Production Databases

Alibaba Bans Anthropic’s Claude Code for Workplace Use

Base44 Rolls Out Base One, Its Specialized Model for Vibe Coding