alchemist-playbook
Evidence-based model-training recipes and run diagnosis (煉丹調參). Use for pretraining, midtraining, SFT, preference tuning, RLVR, LoRA/QLoRA, or speech training; training hyperparameters, loss spikes, data curation, compute budgets, stage evaluation, and experiment reports. Default implementation stack: NVIDIA NeMo (Nemotron recipes and steps, NeMo-RL, NeMo Gym environments); also use when the user mentions NeMo Gym, Nemotron, NeMo-RL, or Megatron-Bridge configs. Pure inference setup or general prompt writing is outside this skill's scope.
Source facts
- Repository
- voidful/AlchemistPlaybook
- Last source activity
- October 27, 2026 at 00:27
- Detected SKILL.md language
- English
- Stars
- 16
- Forks
- 3
Install options
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Review the source files
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Showing SKILL.md
| name | alchemist-playbook |
|---|---|
| description | Evidence-based model-training recipes and run diagnosis (煉丹調參). Use for pretraining, midtraining, SFT, preference tuning, RLVR, LoRA/QLoRA, or speech training; training hyperparameters, loss spikes, data curation, compute budgets, stage evaluation, and experiment reports. Default implementation stack: NVIDIA NeMo (Nemotron recipes and steps, NeMo-RL, NeMo Gym environments); also use when the user mentions NeMo Gym, Nemotron, NeMo-RL, or Megatron-Bridge configs. Pure inference setup or general prompt writing is outside this skill's scope. |