374 skills · sorted by relevance
Fine-tune public CodonFM Encodon checkpoints on labeled coding-sequence or coding-variant data using LoRA, head-only, or full fine-tuning. Use when a user explicitly asks to fine-tune CodonFM or Encodon for regression or classification. Support generic public-v1 Encodon workflows only; reject Decodon, MissenseDataset, missense_synom_agg, and generation workflows.
The English summary is being prepared. Showing the original description.
Stage 4 of the Clinical ASR Flywheel. Use when priority KER is above 0.3 to run stock NeMo SFT on Parakeet TDT v2 and offline cycle N+1 re-eval. NOT for generic word boosting (use /finetune-asr).
The English summary is being prepared. Showing the original description.
Chinese opera playwriting for the qupai (曲牌体) system — 元杂剧, 明清传奇, 昆曲, 川剧高腔, where sung text is filled into fixed tune patterns (填词) rather than built from paired lines varied by metre. Covers stage vs study (李渔 vs 吴梅/王季烈), the 清曲/剧曲 switch, the zaju branch (四折一楔子, one 宫调 and one rhyme per act, 一人主唱, 题目正名), the chuanqi branch (出, 副末开场, 南北合套, 集唐下场诗), tune-filling (格律, 前腔/换头/幺, 衬字, 务头, 尾声, 四十禁), the seventeen 宫调 and their affects, 套数, 排场 alternation, the spoken-to-sung cue, 帮腔, adaptation, and modern practice. From 李渔《闲情偶寄》, 吴梅《顾曲麈谈》《曲学通论》, 王季烈《螾庐曲谈》, 孔尚任's 凡例, 关汉卿, 汤显祖, 魏明伦. Use when writing or revising a 曲牌体 script, choosing a 宫调 for an emotion, laying out a 套数, filling one tune pattern, writing a zaju act or a chuanqi scene, cutting an old 传奇 into a playable version, writing a modern qupai opera, or judging whether a famous script's prosody may be copied. Not for 京剧/豫剧/越剧, which are the banqiang system; see sw-chinese-opera-banqiang.
The English summary is being prepared. Showing the original description.
Iterative skill tuning via execute-evaluate-improve feedback loop. Uses maestro delegate Claude to execute skill, Agy to evaluate quality, and Agent to apply improvements. Iterates until quality threshold or max iterations. Triggers on "skill iter tune", "iterative skill tuning", "tune skill".
The English summary is being prepared. Showing the original description.
Design a hyperparameter tuning strategy for a model — search space, method, and budget. Use when asked to "tune hyperparameters", "define a search space", or "set a tuning budget".
The English summary is being prepared. Showing the original description.
AI relational calibration — reading and adapting to the specific person you are working with. Goes beyond user-intent alignment (solving the right problem) to genuine attunement (meeting the person where they are). Maps communication style, expertise depth, emotional register, and implicit preferences from conversational evidence. Use at the start of a new session, when communication feels mismatched, after receiving unexpected feedback, or when transitioning between very different users or contexts.
The English summary is being prepared. Showing the original description.
中国传统命理分析（八字+紫微斗数）。通过对话收集出生信息，调用排盘脚本获取精确数据， 结合经典典籍进行专业分析。 Triggers: "算命", "算八字", "看八字", "批八字", "排盘", "紫微", "紫微斗数", "帮我看看命", "我的运势", "今年运势", "合婚", "bazi", "ziwei", "fortune"
The English summary is being prepared. Showing the original description.
梅花易数占卜工具，属于 fortune-telling-skills 运势测算套件。基于邵雍《梅花易数》体系， 支持时间起卦、数字起卦、文字起卦等多种起卦方式，通过体用五行生克关系判断吉凶。 起卦为确定性计算 + LLM 断卦解读，无外部 API 依赖。 触发词：梅花易数、起卦、占卜、卦象、体用、八卦占卜、帮我起一卦、算一卦。 不适用于：星座运势、塔罗占卜、八字命理 → 建议使用 fortune-hub。
The English summary is being prepared. Showing the original description.
Fine-tune large language models with LoRA and QLoRA, methods that train small additions to a base model while limiting GPU memory use. Configure training, save task-specific adapters, switch between them, and merge them for deployment.
Tune AgentDB memory use and search speed through quantization, which stores vectors more compactly, search indexes, caching, and batch operations. Includes configuration examples and benchmarks to help balance accuracy, speed, and storage as datasets grow.
Tune AgentDB memory use and search speed through quantization, which stores vectors more compactly, search indexes, caching, and batch operations. Includes configuration examples and benchmarks to help balance accuracy, speed, and storage as datasets grow.
Use Hugging Face Transformers to load pretrained models, generate text, and classify text, images, or audio. Prepare model inputs, control generation settings, and fine-tune models on custom datasets with the Trainer API.
Build searches that find similar meanings using vector databases, which store numerical representations of data. Choose search indexes, combine semantic and keyword search, and tune retrieval quality and response time for search, recommendations, or AI answer systems.
Tune vector indexes, which organize data for similarity searches, to balance speed, memory use, and missed matches. Choose index types, adjust HNSW settings, and assess compression options using benchmarks with real queries.
Export an approved fine-tuned model into a format suited to its hardware and runtime. Choose weight packaging and quantization, which reduces numerical precision, then compare outputs before and after export to check for regressions.
Create rules for Suricata and Snort, systems that detect suspicious network traffic, to flag port scans and reconnaissance. Combine packet patterns with connection counts, tune alert volume, and summarize scanning sources, targets, and activity.
Create and tune Falco rules to detect possible container escapes, where a container gains access to its host. Define detection conditions, alert priorities, and output fields, then refine exceptions to reduce false positives.
Deploy and configure Snort 3 to detect suspicious network traffic. Create custom detection rules, tune community rules to reduce false alerts, test them against captured traffic, and forward alerts to a central security monitoring platform.
Build a repeatable process for Microsoft monthly security updates, from reviewing advisories and ranking vulnerabilities to testing, staged deployment, and verification. Helps teams set response deadlines based on risk and report remaining unpatched systems.
Deploy language models with vLLM through an OpenAI-compatible API or offline batch processing. Tune response speed, throughput, and GPU memory use with quantization, which reduces numerical precision, and execution across multiple GPUs.
Combine capabilities from multiple fine-tuned LLMs without retraining using mergekit. Choose a merging method, configure how models are blended, and evaluate the resulting model on tasks such as math, code generation, and conversation.
Support startup AI strategy by comparing API use, fine-tuning—additional training of an existing model—and in-house development. Assess three-year costs, regulatory risk, self-hosting breakeven points, and hiring priorities to inform leadership decisions.
Prepare to fine-tune open models or Gemini models on Agent Platform. Check the environment, convert and validate training data, choose supported models and settings, and estimate tuning costs for open models.
Deploy Model Garden models to Agent Platform endpoints, the addresses used for model requests. Check supported hardware and estimated hourly costs, inspect deployment progress, test predictions, and remove deployed resources after use.