MLOps
模型部署、评估与运维
simpo-training
magnetoid
Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference model needed, more efficient than DPO. Use for preference alignment when want simpler, faster training than DPO/PPO.
model-eval
molotroka123-cell
Benchmark and validate a local or cloud model for real BOSSMAN roles, measuring speed, memory, capabilities, tool use, context behavior, stability, and cost rather than relying only on public benchmarks.
shap
eyadsibai
Use when "SHAP", "Shapley values", "feature importance", "model explainability", or asking about "explain predictions", "interpretable ML", "feature attribution", "waterfall plot", "beeswarm plot", "model debugging"
experiment-tracking
eyadsibai
Use when "experiment tracking", "MLflow", "Weights & Biases", "wandb", "model registry", "hyperparameter logging", "ML experiments", "training metrics"
llm-training
eyadsibai
Use when "training LLM", "finetuning", "RLHF", "distributed training", "DeepSpeed", "Accelerate", "PyTorch Lightning", "Ray Train", "TRL", "Unsloth", "LoRA training", "flash attention", "gradient checkpointing"
ace-step-inference
SpiralCloudOmega
ACE-Step 1.5 music generation — GGML inference, text-to-music, covers, repainting, CUDA acceleration, 48kHz stereo output for REVITHION STUDIO
scikit-learn
eyadsibai
Use when "scikit-learn", "sklearn", "machine learning", "classification", "regression", "clustering", or asking about "train test split", "cross validation", "hyperparameter tuning", "ML pipeline", "random forest", "SVM", "preprocessing"
hyperparameter-tuning-expert
CMSForNerd
"Analyzes, optimizes, and debugs deep learning hyperparameters including learning rates across PyTorch and TensorFlow."
model-optimization
pluginagentmarketplace
Quantization, pruning, AutoML, hyperparameter tuning, and performance optimization. Use for improving model performance, reducing size, or automated ML.
rtneural-inference
SpiralCloudOmega
Real-time neural network inference for guitar amp/effect modeling using RTNeural C++ library
pymc
eyadsibai
Use when "PyMC", "Bayesian", "MCMC", "probabilistic programming", or asking about "Bayesian regression", "hierarchical model", "NUTS sampler", "posterior distribution", "prior predictive", "credible intervals", "uncertainty quantification"
@tank/opencode-config-mastery
tankpkg
Expert OpenCode configuration for any project. Covers opencode.json schema, MCP server setup (local STDIO, remote HTTP, OAuth), provider and model configuration (75+ providers, custom/local models), plugin management (oh-my-opencode, morph, wakatime), agent wiring in config, custom slash commands, rules/instructions files, keybinds, permission system, and security best practices for MCP servers. Includes trusted MCP server catalog and context budget management strategies. Synthesizes OpenCode source (sst/opencode), official docs, MCP specification, OWASP MCP security guidelines, and community ecosystem analysis. Trigger phrases: "opencode.json", "opencode config", "configure opencode", "MCP server", "mcp setup", "add mcp", "mcp config", "mcp oauth", "opencode provider", "add provider", "custom provider", "local model", "ollama opencode", "LM Studio", "opencode plugin", "oh-my-opencode", "opencode permission", "permission config", "bash permission", "slash command", "custom command", "opencode command", ".opencode", "instructions file", "rules file", "AGENTS.md", ".cursorrules", "opencode keybind", "keybinds", "context budget", "too many tools", "MCP security", "tool poisoning", "opencode setup", "configure model", "small_model", "enabled_providers", "disabled_providers", "opencode agent config", "watcher ignore", "opencode share", "opencode theme", "opencode lsp", "opencode formatter", "variable substitution", "{env:}", "{file:}", "/connect", "auth.json", "mcp-auth.json"
mle-workflow
anavvanzin
Production machine-learning engineering workflow for data contracts, reproducible training, model evaluation, deployment, monitoring, and rollback. Use when building, reviewing, or hardening ML systems beyond one-off notebooks.
transformers
eyadsibai
Use when "HuggingFace Transformers", "pre-trained models", "pipeline API", or asking about "text generation", "text classification", "question answering", "NER", "fine-tuning transformers", "AutoModel", "Trainer API"
ns-principle-outcome-oriented-execution
noahdgrant
"Apply during planned rewrites and migrations with explicit phase boundaries. Converge on the target architecture; don't preserve smooth intermediate states with throwaway compatibility code."
manuscript-writing
BasileChretien
Draft or revise manuscript prose that reads as though a person wrote it. Use when writing any section, when check reports ai-phrasing, ai-cadence or vague-attribution, or when revising text that was drafted quickly.
multimodal-models
eyadsibai
Use when "CLIP", "Whisper", "Stable Diffusion", "SDXL", "speech-to-text", "text-to-image", "image generation", "transcription", "zero-shot classification", "image-text similarity", "inpainting", "ControlNet"
Sak-family-auto-cycle
beer-sakthai
Dispatch all six Sak Family personas (SakKing/SakThai/SakSee/SakSit/SakTan/SakJules) as one parallel fan-out, each sustaining up to 3 Dream-through-Growth cycle rounds. Use whenever the user asks to run the family auto-cycle, run the Sak Family, get the personas working together, dispatch them as a team, or kick off a family round — even if they name only "the family" or "the cycle" without listing the personas.
SakJules-SakThai-ecosystem-improvement-cron
beer-sakthai
"One-shot cron task that improves one HF asset per run — model card, cross-links, dataset integrity, or promotion."
peft-fine-tuning
magnetoid
Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models (7B-70B) with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or for multi-adapter serving. HuggingFace's official library integrated with transformers ecosystem.
data-science
eyadsibai
Use when "statistical modeling", "A/B testing", "experiment design", "causal inference", "predictive modeling", or asking about "hypothesis testing", "feature engineering", "data analysis", "pandas", "scikit-learn"
ns-principle-model-the-domain
noahdgrant
"Apply when writing stateful logic, or when code branches a lot or repeats a shape assumption across files. Encode the domain in a structure instead of scattered conditionals."
SakJules-bigquery-bigframes
beer-sakthai
">- Generates Python code using BigQuery DataFrames (BigFrames), the pandas/scikit-learn-style API over BigQuery. Use when writing BigFrames code or doing pandas-style dataframe/ML work against BigQuery (e.g. in a notebook). Dont use for SQL-first w"
hugging-face-evaluation-manager
Nymbo
Add and manage evaluation results in Hugging Face model cards. Supports extracting eval tables from README content, importing scores from Artificial Analysis API, and running custom model evaluations with vLLM/lighteval. Works with the model-index metadata format.