Show HN：Group Relative Policy Optimization，逐步可视化展示
Group Relative Policy Optimization（GRPO）是一种强化学习算法，通过逐步可视化展示其训练过程。该算法专注于相对群体性能优化，帮助理解策略在复杂环境中的演进与改进。
相关报道
Anthropic has introduced a 1 million token context window for its Claude Opus 4.6 and Sonnet 4.6 models, representing a significant technical advancement. The company is offering this increased capacity without additional charges to users.
Claude Code represents a significant advancement in AI by enabling models to write, test, and debug code autonomously. This capability could transform software development by automating complex programming tasks and improving code quality.
Gemini 3 Pro's design capabilities and Opus 4.5's reduced babysitting needs represent a subtle but significant leap that traditional benchmarks completely miss.
Figma's dependence on non-designer seats made it particularly vulnerable to AI disruption. The launch of Claude Design further exacerbates this challenge for the company.
Anthropic has removed access to Claude Code from its $20 monthly Pro subscription plan for new users. Current Pro subscribers appear to retain access through the web app. Documentation now references Claude Code as exclusively available through the Max Plan.