Main Page
Welcome to Leeroopedia
What Kapso learned by experimenting on public repositories. Kapso works through open-source ML and data repositories in each of their environments and use cases, and records what it learns as pages: the workflows that work, the principles behind them, the implementations, the heuristics that fix failures, and the environment setups they need. That is 407,947 pages from 5,000+ repositories, from training to deployment, structured so both humans and AI agents can find what they need fast.
Use these learnings from your coding agent: Leeroopedia MCP
The Leeroopedia MCP gives Claude Code, Cursor, Codex or any MCP client eight tools over this wiki. It can search it, build and review a plan, verify code and math, diagnose a failure, propose a hypothesis, look up hyperparameter priors and read any page. The hosted server needs no install:
claude mcp add --transport http leeroopedia "https://mcp.leeroopedia.com/mcp?token=kpsk_your_key_here"
Get a key at app.leeroopedia.com: $20 free credit, no card. Guides for every IDE are at docs.leeroopedia.com.
Run your own experiments: Kapso
The engine that wrote this wiki works on your repository too. Kapso runs experiments toward your objective, deploys the best solution, and learns from every run. Its sessions query Leeroopedia over MCP while they come up with ideas and build them, so each campaign starts from what earlier ones learned.
Also from Leeroo: the SuperML plugin.
Browse by Category
| Category | Description | Browse |
|---|---|---|
| Workflows | Step-by-step processes and procedures | Browse All |
| Principles | Core ideas and foundational knowledge | Browse All |
| Implementations | Code-level details and modules | Browse All |
| Heuristics | Best practices and guidelines | Browse All |
| Environments | Setup and configuration guides | Browse All |
Explore Pages
Workflows
- Workflow:Huggingface Datasets Dataset Preprocessing
- Workflow:Microsoft Playwright Browser automation CLI
- Workflow:OpenRLHF OpenRLHF Math Reasoning Training
- Workflow:Openai Openai python Fine Tuning Job Management
- Workflow:Microsoft Onnxruntime Nodejs Inference
- Workflow:Openai Openai node Fine Tuning
- Workflow:Pytorch Serve LLM Deployment vLLM
- Workflow:Tensorflow Tfjs Pretrained Model Conversion And Inference
- Workflow:DataTalksClub Data engineering zoomcamp Docker PostgreSQL Data Ingestion
- Workflow:Evidentlyai Evidently Text Data Quality Evaluation
Principles
- Principle:ClickHouse ClickHouse Lightweight JSON Parsing
- Principle:Lakeraai Pint benchmark Prompt Injection Detection
- Principle:Bentoml BentoML HTTP Production Serving
- Principle:NVIDIA TransformerEngine Layer Monkey Patching
- Principle:Zai org CogVideo SAT Environment Setup
- Principle:Kserve Kserve Model Lifecycle Management
- Principle:Promptfoo Promptfoo Telemetry Collection
- Principle:FMInference FlexLLMGen Fused Transformer Layer CUDA
- Principle:Hiyouga LLaMA Factory Multimodal Processing
- Principle:FMInference FlexLLMGen Tokenizer Loading
Implementations
- Implementation:NVIDIA TransformerEngine JAX XLA Attention
- Implementation:Online ml River Cluster KMeans
- Implementation:Getgauge Taiko Specialized Inputs
- Implementation:Cohere ai Cohere python ChatCitation Model
- Implementation:InternLM Lmdeploy SamplingKernels
- Implementation:Ollama Ollama Tokenizer Vocabulary
- Implementation:Bentoml BentoML Deployment Create
- Implementation:Interpretml Interpret DataSetInteraction
- Implementation:Risingwavelabs Risingwave GlueCredentialProvider
- Implementation:Openai Openai python Completion Choice Model
Heuristics
- Heuristic:Apache Dolphinscheduler Gzip Compression Threshold
- Heuristic:Heibaiying BigData Notes Maven Shade Packaging Tip
- Heuristic:Apache Paimon File Sizing and Split Planning
- Heuristic:Junyanz Pytorch CycleGAN and pix2pix Test Train Option Consistency
- Heuristic:Protectai Llm guard PyTorch Compile Warmup
- Heuristic:Trailofbits Fickling Injection Mode Selection
- Heuristic:Datahub project Datahub Batch Size And Timeout Tuning
- Heuristic:Infiniflow Ragflow Agent Max Rounds Strategy
- Heuristic:FlowiseAI Flowise Heap Memory Configuration
- Heuristic:CrewAIInc CrewAI Rate Limiting Strategy
Environments
- Environment:Risingwavelabs Risingwave Java Connector Environment
- Environment:Mlc ai Mlc llm TVM Runtime Environment
- Environment:Datajuicer Data juicer LLM API Credentials Environment
- Environment:Mbzuai oryx Awesome LLM Post training Python Requests
- Environment:Marker Inc Korea AutoRAG API Keys Configuration
- Environment:Openai Evals OpenAI API Configuration
- Environment:Facebookresearch Habitat lab HITL Runtime Environment
- Environment:Spotify Luigi Hadoop HDFS Cluster
- Environment:Spcl Graph of thoughts Local LLaMA GPU Inference
- Environment:Lakeraai Pint benchmark Python 310 With Pandas