OpenSkillIndex← back to search
claude-code

AutoResearchClaw mixed-precision

Use FP16/BF16 mixed precision to accelerate training and reduce memory. Use when optimizing GPU performance.

install
source · Clone the upstream repo
git clone https://github.com/aiming-lab/AutoResearchClaw
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/aiming-lab/AutoResearchClaw "$T" && mkdir -p ~/.claude/skills && cp -r "$T/researchclaw/skills/builtin/tooling/mixed-precision" ~/.claude/skills/aiming-lab-autoresearchclaw-mixed-precision && rm -rf "$T"
manifest: researchclaw/skills/builtin/tooling/mixed-precision/SKILL.md
source content

Mixed Precision Training Best Practice

Use torch.cuda.amp for automatic mixed precision:

  • Wrap forward pass in torch.cuda.amp.autocast()
  • Use GradScaler for loss scaling
  • BF16 preferred over FP16 on Ampere+ GPUs (RTX 3xxx, A100, RTX 4xxx)
  • Watch for NaN gradients — reduce learning rate if needed
  • Do NOT use amp with custom CUDA kernels unless tested