Inference Spec Decode Tune

Tune a speculative-decoding DRAFT head's TRAINING hyperparameters (global batch, learning rate, accumulation, warmup) for a target LLM, optimizing the TRUE serving in-engine acceptance length (vLLM spec_decode counters) with a cheap training-acc proxy for triage. Reuses the search ALGORITHMS (hyperband/grid/random natively. Bayesian-TPE via optuna if installed) wired to a GB300/managed K8s pod launcher + an offline EAGLE3/DFlash trainer (distinct from the MLPerf ai_tuning contract, which does not map to draft training). The draft-training analog of inference-tune-sweep (which tunes vLLM SERVING config). Triggers on "tune the draft head", "tune eagle3 / dflash training", "bayesian/hyperband tune the draft", "search global batch and LR for the speculator", "spec-decode hyperparameter sweep", "draft-training tuner", or any combination of "tune / sweep / optimize / search / bayesian / hyperband" with "eagle3 / dflash / draft / speculator / spec-decode".

mkdir -p ~/.claude/skills/inference-spec-decode-tune
curl -L https://claudskills.com/skills/inference-spec-decode-tune/SKILL.md \
  -o ~/.claude/skills/inference-spec-decode-tune/SKILL.md

@misc{inference-spec-decode-tune-2026,
  author    = {cfregly},
  title     = {Inference Spec Decode Tune [Claude Code skill]},
  year      = {2026},
  publisher = {ClaudSkills},
  url       = {https://claudskills.com/skills/inference-spec-decode-tune/}
}