#Group Relative Policy Optimization
89 related articles
·7 min
Dispersion Loss: A Key Technique for Solving Embedding Condensation in Small Language Models
Embedding condensation is a hidden bottleneck in small language model training. Dispersion Loss combats this by enforcing representation spread during training at zero inference cost.
Read more →
·7 min
Claude Code Command Orchestration: One-Click Content Research and Auto-Publishing
Learn how to orchestrate Claude Code custom commands to chain content research and social media publishing agents into a fully automated workflow with one command.
Read more →
·6 min
Anthropic Launches Life Sciences Hackathon with $100K to Empower AI-Driven Research
Anthropic launches a global life sciences hackathon with Gladstone Institutes, offering $100K in API credits to advance AI-driven scientific research using Claude Science and Claude Code.
Read more →
·7 min
AI Agent Practical Guide: From Content Generation to Enterprise Applications
A deep dive into AI Agent's two core directions: 2C content generation (text/images/video) and 2B enterprise applications (RAG/AutoGen/LLM integration). With real startup cases and practical methods.
Read more →
·8 min
Free Top-Tier AI Models on Aggregator Platforms? A Sober Analysis and Pitfall Avoidance Guide
In-depth analysis of AI aggregator platforms claiming free access to GPT, DeepSeek, and Gemini. Reveals hidden data risks, business logic, and recommends legitimate alternatives like OpenRouter and Poe.
Read more →
·4 min
LLM Job Interviews: Multi-Agent and Harness Engineering Are Now Must-Know Topics
Deep analysis of LLM job interview essentials: Multi-Agent architecture, Harness engineering, Agent Loop, sandbox isolation, and memory management with career transition tips.
Read more →
·3 min
Sakana AI Releases Fugu Ultra: How Model Orchestration Achieves Frontier AI Performance
Sakana AI releases Fugu Ultra, achieving frontier AI performance through autonomous model orchestration. Deep dive into its technology, strategic implications, and impact on global AI competition.
Read more →
·3 min
Microsoft Copilot Cowork Introduces DeepSeek: Multi-Model Architecture and Enterprise AI Agent Strategy Explained
Microsoft Copilot Cowork launches with multi-model architecture, considering DeepSeek V4 as a low-cost option. Deep dive into usage-based pricing, WebIQ search, and Microsoft's enterprise AI agent strategy.
Read more →
·3 min
Why SFT Can't Fix the Root Cause of JSON Errors: How GRPO Correctness Training Breaks Through Coding Agent Bottlenecks
Analysis of why SFT can't fix coding agent JSON errors and how GRPO's binary reward signals and synchronized weight updates train directly for correctness.
Read more →
Tech Frontiers·3 min
GitHub Agent HQ Launch: AI Coding Tools Enter the Era of Platform Competition
GitHub Universe unveils Agent HQ platform for unified coding agent management, Copilot upgrades with multi-model support. OpenAI completes restructuring, Anthropic tests new model, NVIDIA open-sources AI models.
Read more →
Deep Dives·3 min
Complete Guide to LLM Training: Pre-training, SFT Fine-tuning, and Preference Alignment Explained
Complete guide to the three core LLM training stages: pre-training, supervised fine-tuning (SFT), and preference alignment (DPO/PPO), covering LoRA, distillation, quantization, and pruning.
Read more →
Product Reviews·2 min
Hands-On Review of a High-Freedom AI Companion Chat App: Unrestricted Roleplay Conversation Experience
Hands-on review of an AI companion chat app featuring unrestricted dialogue, analyzing its roleplay, story interaction, and custom character creation while comparing it to Character.AI and SillyTavern.
Read more →
Research·3 min
Inside Cursor Composer 2 Training: A Complete Breakdown of Its Distributed Reinforcement Learning Architecture
Deep dive into how Cursor trained Composer 2 on Fireworks: async pipeline architecture, MoE numerical precision challenges, Router Replay, and global distributed GPU coordination.
Read more →
Tutorials·2 min
ComfyUI Video Matting Tutorial: Complete Workflow with RMBG Model + APNG Export
Step-by-step ComfyUI video matting workflow using the RMBG model for AI background removal, generating transparent APNG animations compressed from 160MB to 30MB for mobile use.
Read more →
Research·2 min
Technical Breakdown of Cursor Composer 2's Distributed RL Training
Deep dive into how Cursor trained Composer 2 via distributed RL, covering async pipelines, MoE numerical alignment, global weight sync, and more.
Read more →
Tutorials·2 min
DeepSeek R1 Usage Guide: Prompt Techniques & Practical Tips (2025)
A detailed guide to DeepSeek R1's reasoning capabilities and prompt techniques, covering reasoning vs. general models, core capabilities, and practical prompt strategies for 2025.
Read more →
Tutorials·4 min
Save Windsurf Credits: MCP Plugin Extends Each Conversation by 5–6 Extra Interactions
How to save Windsurf credits using an MCP plugin that extends each conversation by 5–6 interactions. Covers setup, core principles, ban risks, and security tips.
Read more →
Product Reviews·3 min
Trae + Gemini for HarmonyOS Incoming Call Simulator: 70% AI-Generated Code, Completed in 5 Days
Developer uses Trae + Gemini 3 Pro to build a HarmonyOS incoming call simulator with 70% AI-generated code, compressing 10 days of work to 5. Covers four core modules, AI limitations, and practical tips.
Read more →
Deep Dives·2 min
DeepSeek V3.2 Deep Dive: Three Major Breakthroughs in DSA Sparse Attention, Reinforcement Learning, and Agent Capabilities
Deep analysis of DeepSeek V3.2 and V3.2 Special: DSA sparse attention for faster long-context processing, RL compute at 10% of pre-training, and Agent task synthesis across 1,800 environments.
Read more →
Tutorials·3 min
OpenAI Open-Sources GPT-OSS: Run O4-Level Models on 16GB VRAM — Complete Deployment Guide
OpenAI open-sources GPT-OSS (20B/120B) with MOE architecture and native FP4 precision. Run O3-level reasoning on a single RTX 4090. Full deployment guide for Ollama, vLLM, and more.
Read more →