Implement GRPO (Group Relative Policy Optimization) fine-tuning for vision-language models on small datasets. Use when SFT underperforms or training data is limited (<1000 examples).
安装
npx skills add aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock --skill grpo-finetuningnpx skills add aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock --skill grpo-finetuning -a claude-code -gnpx skills add aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock --skill grpo-finetuning -a codex -ggemini skills install https://github.com/aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock.git --path assets/claude-code-plugins/plugins/ml-training/skills/grpo-finetuninggh skill install aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock grpo-finetuninggit clone --depth 1 https://github.com/aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock /tmp/aws-solutions-library-samples__guidance-for-claude-code-with-amazon-bedrock
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/aws-solutions-library-samples__guidance-for-claude-code-with-amazon-bedrock/assets/claude-code-plugins/plugins/ml-training/skills/grpo-finetuning ~/.claude/skills/grpo-finetuning # Claude Code
cp -R /tmp/aws-solutions-library-samples__guidance-for-claude-code-with-amazon-bedrock/assets/claude-code-plugins/plugins/ml-training/skills/grpo-finetuning ~/.agents/skills/grpo-finetuning # Codex, Cursor, Gemini CLI, Copilot, OpenCode