Skills for your AI assistant
Add a skill and Claude Code, Codex, Cursor or a similar assistant learns one more job.
Fine-tuning and quantization
10 skillsFine-tuning, post-training, quantizing and compressing models.
- fine-tuning-expert4.6k installsHelps fine-tune language models with LoRA adapters, training datasets, hyperparameters and deployment.
- codonfm-finetune15 installsFine-tunes public CodonFM Encodon checkpoints on labeled coding-sequence data for regression or classification.
- kermt-pretrain-scratch12 installsPretrains a brand-new KERMT model from scratch on your molecule corpus.
- bad-case-to-dpoTurns failed production runs into training data, from failure analysis and preference pairs to LoRA training and validation.
- curly-quote-sftSuggests a skill about training a model on curly-quote text, based on its name only.
- kermt-add-cmim-pretrain10 installsConverts a KERMT checkpoint to hybrid and continues pretraining on your molecule corpus.
- kermt-continue-pretrain10 installsContinues KERMT pretraining on your own SMILES molecule corpus from an existing checkpoint.
- kermt-finetune10 installsFine-tunes a pretrained KERMT encoder on your labeled CSV of molecules.
- post-training-strategyGuides choosing and running post-training for agent models, from mid-training and SFT to RL, DPO and distillation.
- reward-designHelps you design reward functions for RL training and avoid reward hacking.