Train GPT-2 scale models (~124M parameters) efficiently on a single GPU. Covers GPT-124M architecture, tokenized dataset loading (e.g., HuggingFace Hub shards), modern optimizers (Muon, AdamW), mixed precision training, and training loop implementation.
安装
npx skills add benchflow-ai/skillsbench --skill nanogpt-trainingnpx skills add benchflow-ai/skillsbench --skill nanogpt-training -a claude-code -gnpx skills add benchflow-ai/skillsbench --skill nanogpt-training -a codex -ggemini skills install https://github.com/benchflow-ai/skillsbench.git --path analysis/tasks_skills_low/mhc-layer-impl/level_3/nanogpt-traininggh skill install benchflow-ai/skillsbench nanogpt-traininggit clone --depth 1 https://github.com/benchflow-ai/skillsbench /tmp/benchflow-ai__skillsbench
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/benchflow-ai__skillsbench/analysis/tasks_skills_low/mhc-layer-impl/level_3/nanogpt-training ~/.claude/skills/nanogpt-training # Claude Code
cp -R /tmp/benchflow-ai__skillsbench/analysis/tasks_skills_low/mhc-layer-impl/level_3/nanogpt-training ~/.agents/skills/nanogpt-training # Codex, Cursor, Gemini CLI, Copilot, OpenCode