Python
Prototype an offline audio-analysis pipeline with arrays and functions; separate file I/O from analysis so each can be tested.
1 introductory training example · not runtime-testedExplore with Keeks ↗KEEKS'S LEARNING MAP
Libraries & tools.
From a single sample to a shader, a model, or a live performance. Explore the 101 technologies from Kiki's notes.
Dataset prepared · GPU fine-tuning pending
These are topics represented in training examples, not installed integrations or verified model abilities. The demo retrieves authored notes. No trained adapter is connected.
101 new introductory examples · 233 total training examples · 25 validation examples · 185 demo notes.
One primer per tool is an introduction, not comprehensive API coverage. The new primers are original synthetic text, await expert review, and do not import full documentation. The existing 25 evaluation prompts do not evaluate every tool.
101 tools
Languages & web
Java
Build a typed MIDI-event model; distinguish managed application code from a hard real-time audio callback.
1 introductory training example · not runtime-testedExplore with Keeks ↗C++
Implement a sample-processing loop with preallocated buffers; avoid allocation and blocking locks in the audio callback.
1 introductory training example · not runtime-testedExplore with Keeks ↗C#
Build a .NET control interface for an audio service; keep UI work off the audio-processing thread.
1 introductory training example · not runtime-testedExplore with Keeks ↗OCaml
Represent musical events with algebraic data types and pattern matching; make invalid event states explicit.
1 introductory training example · not runtime-testedExplore with Keeks ↗JavaScript
Connect browser controls to an audio graph; distinguish event-loop timing from audio-clock scheduling.
1 introductory training example · not runtime-testedExplore with Keeks ↗MATLAB
Explore an FFT and plot its frequency axis in Hz; record sample rate, transform size and window.
1 introductory training example · not runtime-testedExplore with Keeks ↗R
Analyze repeated audio experiments with tidy tables and uncertainty estimates; distinguish observations from independent subjects.
1 introductory training example · not runtime-testedExplore with Keeks ↗SQL
Store take metadata and query recordings by sample rate; parameterize user input rather than concatenating SQL.
1 introductory training example · not runtime-testedExplore with Keeks ↗HTML
Structure a chatbot with labels, headings, and navigation; use semantic controls rather than clickable divs.
1 introductory training example · not runtime-testedExplore with Keeks ↗CSS
Style responsive controls and readable code blocks; preserve keyboard focus and contrast.
1 introductory training example · not runtime-testedExplore with Keeks ↗GLSL
Write a fragment shader that maps coordinates to color; verify shader version and coordinate conventions.
1 introductory training example · not runtime-testedExplore with Keeks ↗Swift
Build a native Apple interface for an audio experiment; separate UI state from audio-engine state.
1 introductory training example · not runtime-testedExplore with Keeks ↗Python, audio & science
pydub
Edit and concatenate offline audio segments; check format decoding dependencies and do not confuse milliseconds with sample indices.
1 introductory training example · not runtime-testedExplore with Keeks ↗librosa
Extract music and audio features; verify sample rate and mono settings at loading and keep the feature time axis explicit.
1 introductory training example · not runtime-testedExplore with Keeks ↗soundfile
Read and write audio arrays using libsndfile; preserve sample rate, channel shape, and output subtype.
1 introductory training example · not runtime-testedExplore with Keeks ↗pygame
Prototype a visual rhythm game with an event loop; keep simulation time independent of frame rate.
1 introductory training example · not runtime-testedExplore with Keeks ↗pygame.mixer
Play sample-based sounds in a pygame application; mixer buffer size trades responsiveness against underrun tolerance.
1 introductory training example · not runtime-testedExplore with Keeks ↗Pillow (PIL)
Load, resize, and composite raster images; distinguish RGB from RGBA and preserve alpha deliberately.
1 introductory training example · not runtime-testedExplore with Keeks ↗OpenGL
Render geometry with GPU buffers and shaders; an OpenGL context is required and calls must match its supported version.
1 introductory training example · not runtime-testedExplore with Keeks ↗matplotlib
Plot waveform and spectrum diagnostics; label units and distinguish linear magnitude from decibels.
1 introductory training example · not runtime-testedExplore with Keeks ↗OpenCV
Process camera frames for interactive control; verify BGR/RGB order and capture timestamps before measuring latency.
1 introductory training example · not runtime-testedExplore with Keeks ↗NumPy
Represent samples as arrays and compute numerical transforms; choose axis and dtype explicitly for multichannel audio.
1 introductory training example · not runtime-testedExplore with Keeks ↗Pandas
Join recording metadata and feature tables; validate join keys to avoid accidentally duplicating observations.
1 introductory training example · not runtime-testedExplore with Keeks ↗scikit-learn
Train a baseline classifier on audio features; split by recording or subject before fitting preprocessing to avoid leakage.
1 introductory training example · not runtime-testedExplore with Keeks ↗MNE
Analyze neurophysiological recordings such as EEG or MEG; preserve channel metadata, units and event timing. This is research analysis, not a diagnosis.
1 introductory training example · not runtime-testedExplore with Keeks ↗PsychoPy
Design a controlled perception experiment; measure actual display and audio timing on the target hardware.
1 introductory training example · not runtime-testedExplore with Keeks ↗MediaPipe
Estimate landmarks from camera frames for musical control; smooth noisy coordinates and handle missing detections.
1 introductory training example · not runtime-testedExplore with Keeks ↗AI & model training
PyTorch
Build tensor-based models and training loops; separate training and evaluation modes and manage device placement explicitly.
1 introductory training example · not runtime-testedExplore with Keeks ↗TensorFlow
Build and train tensor-based models; verify shape, dtype and preprocessing consistency between training and inference.
1 introductory training example · not runtime-testedExplore with Keeks ↗Hugging Face Transformers
Load transformer architectures and tokenizers; match the model revision and native chat template across training and inference.
1 introductory training example · not runtime-testedExplore with Keeks ↗Hugging Face datasets
Load and transform training records; keep provenance and group-based split rules through mapping operations.
1 introductory training example · not runtime-testedExplore with Keeks ↗MLflow
Track model experiments, parameters, and artifacts; log dataset and code versions alongside metrics.
1 introductory training example · not runtime-testedExplore with Keeks ↗Trainer
Use the Hugging Face training loop abstraction; check labels and data collator behavior rather than assuming response-only loss is automatic.
1 introductory training example · not runtime-testedExplore with Keeks ↗AutoTokenizer
Load the tokenizer associated with a model; inspect pad tokens, chat formatting and truncation behavior.
1 introductory training example · not runtime-testedExplore with Keeks ↗AutoModelForCausalLM
Load an autoregressive language model architecture; model loading alone does not fine-tune it or create a chat template.
1 introductory training example · not runtime-testedExplore with Keeks ↗Jupyter
Explore data and experiments in notebooks; restart and run all cells to expose hidden execution-order dependencies.
1 introductory training example · not runtime-testedExplore with Keeks ↗Ollama
Run supported language models locally; inference with an imported model is distinct from training a new adapter.
1 introductory training example · not runtime-testedExplore with Keeks ↗DeepSeek
Treat this as a model family or provider name, not one fixed architecture; request the exact model ID, revision and license before choosing a training recipe.
1 introductory training example · not runtime-testedExplore with Keeks ↗WizardCoder
Treat this as a code-model family; inspect the exact checkpoint, base model, and license instead of assuming all variants share one configuration.
1 introductory training example · not runtime-testedExplore with Keeks ↗Vicuna
Treat this as a chat-model family; verify checkpoint-specific conversation formatting and use terms before deployment.
1 introductory training example · not runtime-testedExplore with Keeks ↗Extraction & documents
BeautifulSoup (bs4)
Parse HTML and select document elements; it does not execute page JavaScript or fetch pages by itself.
1 introductory training example · not runtime-testedExplore with Keeks ↗Scrapy
Organize crawls with spiders and item pipelines; respect access rules, throttle requests and retain source URLs.
1 introductory training example · not runtime-testedExplore with Keeks ↗Readability-lxml
Extract main article content from HTML; inspect outputs because navigation removal can also discard useful code or tables.
1 introductory training example · not runtime-testedExplore with Keeks ↗requests
Fetch HTTP resources from Python; set timeouts, handle status codes and keep credentials out of logs.
1 introductory training example · not runtime-testedExplore with Keeks ↗PyPDF2
Work with PDF pages and embedded text; scanned image pages need a separate OCR stage, and extraction order may differ from reading order.
1 introductory training example · not runtime-testedExplore with Keeks ↗pdfminer
Extract text and layout from PDFs; inspect columns, ligatures and code indentation before creating training examples.
1 introductory training example · not runtime-testedExplore with Keeks ↗Creative graphics
TouchDesigner
Connect CHOP, TOP and other operators for interactive audiovisual work; verify whether control channels and audio channels use compatible rates.
1 introductory training example · not runtime-testedExplore with Keeks ↗Hydra
Compose live-coded visual transformations; keep browser performance and audio-input permissions separate from shader syntax.
1 introductory training example · not runtime-testedExplore with Keeks ↗Processing
Sketch interactive graphics with a draw loop; distinguish the Processing Java environment from p5.js in a browser.
1 introductory training example · not runtime-testedExplore with Keeks ↗p5.js
Build browser sketches with setup and draw; separate audio timing from animation frame timing.
1 introductory training example · not runtime-testedExplore with Keeks ↗Three.js
Construct a browser 3D scene with cameras, geometry and materials; dispose GPU resources when replacing scenes.
1 introductory training example · not runtime-testedExplore with Keeks ↗D3.js
Bind structured data to visual elements; choose scales and axes that accurately represent units and domains.
1 introductory training example · not runtime-testedExplore with Keeks ↗WebGL
Render through a browser GPU API; check context support and handle context loss rather than assuming a persistent renderer.
1 introductory training example · not runtime-testedExplore with Keeks ↗OpenFrameworks (ofx)
Build C++ creative applications with openFrameworks; ofx names typically identify addons whose compatibility must be checked separately.
1 introductory training example · not runtime-testedExplore with Keeks ↗Blender
Model and animate 3D scenes; distinguish rendering, geometry processing and scripting workloads when planning performance.
1 introductory training example · not runtime-testedExplore with Keeks ↗Houdini
Build procedural geometry networks; cache expensive stages and record parameters so results can be reproduced.
1 introductory training example · not runtime-testedExplore with Keeks ↗parallax.js
Add depth-like pointer or motion effects; confirm the exact package because similarly named projects have different APIs.
1 introductory training example · not runtime-testedExplore with Keeks ↗Web MIDI
Exchange MIDI messages with browser-accessible devices; feature-detect support and handle permission denial. MIDI events do not carry recorded audio.
1 introductory training example · not runtime-testedExplore with Keeks ↗MediaPipe TouchDesigner plugin
Bridge landmark tracking into TouchDesigner; request the exact plugin repository and release before giving component-specific setup instructions.
1 introductory training example · not runtime-testedExplore with Keeks ↗Live coding & music
TidalCycles
Describe musical patterns evaluated by a live-coding environment; confirm the sound engine and sample bank rather than assuming every sound name exists.
1 introductory training example · not runtime-testedExplore with Keeks ↗Strudel
Use a browser-based pattern environment for musical sequences; check sample availability and browser audio activation before debugging rhythm.
1 introductory training example · not runtime-testedExplore with Keeks ↗Gibber
Create live-coded musical sequences and visuals; identify the environment version before using instrument-specific syntax.
1 introductory training example · not runtime-testedExplore with Keeks ↗Glicol
Describe audio graphs with a live-coding language; check the active engine and available nodes before composing a patch.
1 introductory training example · not runtime-testedExplore with Keeks ↗Shadertoy
Prototype fragment-shader visuals in its host environment; host-provided uniforms and entry points need adaptation outside Shadertoy.
1 introductory training example · not runtime-testedExplore with Keeks ↗GLSL Sandbox
Explore fragment shaders in a browser playground; verify its uniforms and GLSL dialect rather than pasting another host's shader unchanged.
1 introductory training example · not runtime-testedExplore with Keeks ↗p5live
Use a live-coding environment around p5.js; confirm the editor version and shared session behavior before collaborating.
1 introductory training example · not runtime-testedExplore with Keeks ↗OpenProcessing
Share and explore creative sketches; inspect each sketch's license and dependencies before reusing code in a dataset.
1 introductory training example · not runtime-testedExplore with Keeks ↗Ableton Live
Arrange and process musical clips and tracks; measure end-to-end latency with the chosen interface and buffer configuration.
1 introductory training example · not runtime-testedExplore with Keeks ↗Max for Live
Build Max devices inside Ableton Live; distinguish message-rate controls from signal-rate audio processing.
1 introductory training example · not runtime-testedExplore with Keeks ↗Logic Pro
Record, arrange and mix audio and MIDI; preserve project sample rate and distinguish software monitoring from direct monitoring.
1 introductory training example · not runtime-testedExplore with Keeks ↗Kontakt
Create sample-based instruments; confirm the instrument edition and scripting support before planning automation.
1 introductory training example · not runtime-testedExplore with Keeks ↗TAL-Sampler
Explore sampler-based synthesis and resampling character; compare matched output levels when evaluating sonic differences.
1 introductory training example · not runtime-testedExplore with Keeks ↗Jam Origin MIDI Guitar 2
Convert guitar performance into MIDI control; tracking depends on input quality, pitch content and settings, so measure latency on your own setup.
1 introductory training example · not runtime-testedExplore with Keeks ↗Apps & deployment
Gradio
Build a quick model-demo interface; keep model execution and secrets on the server and bound user requests.
1 introductory training example · not runtime-testedExplore with Keeks ↗Streamlit
Build an interactive Python data app; understand reruns and cache lifetime before loading a large model.
1 introductory training example · not runtime-testedExplore with Keeks ↗FastAPI
Expose validated Python HTTP endpoints; bound concurrent GPU requests and avoid blocking the event loop with synchronous work.
1 introductory training example · not runtime-testedExplore with Keeks ↗Node.js
Run server-side JavaScript and tooling; move heavy CPU work away from the main event loop.
1 introductory training example · not runtime-testedExplore with Keeks ↗Flutter
Build interfaces with Dart across supported platforms; check audio-plugin capabilities on each target platform.
1 introductory training example · not runtime-testedExplore with Keeks ↗React Native
Build native mobile interfaces with React; test native audio and camera modules separately on each operating system.
1 introductory training example · not runtime-testedExplore with Keeks ↗WebSocket
Maintain a bidirectional connection for live controls; include timestamps and reconnect handling rather than assuming delivery timing is constant.
1 introductory training example · not runtime-testedExplore with Keeks ↗MQTT
Publish and subscribe to device messages through a broker; choose QoS deliberately and do not treat broker delivery as hard real-time scheduling.
1 introductory training example · not runtime-testedExplore with Keeks ↗ONNX
Represent supported model graphs for interchange; compare exported outputs against the original model on representative inputs.
1 introductory training example · not runtime-testedExplore with Keeks ↗TensorRT
Optimize supported models for NVIDIA inference; verify precision, dynamic shapes and numerical accuracy on the target GPU.
1 introductory training example · not runtime-testedExplore with Keeks ↗Developer tools & integrations
Conda
Manage environments and package dependencies; export the environment used for an experiment rather than relying on the global installation.
1 introductory training example · not runtime-testedExplore with Keeks ↗Anaconda
Use a Python distribution and ecosystem of tools; distinguish the distribution from Conda, the environment manager.
1 introductory training example · not runtime-testedExplore with Keeks ↗Git
Track source changes with commits; keep large datasets and credentials out of ordinary source history.
1 introductory training example · not runtime-testedExplore with Keeks ↗GitHub
Host repositories and collaboration workflows; a published repository or passing workflow does not prove a model was trained.
1 introductory training example · not runtime-testedExplore with Keeks ↗Docker
Package an application environment in containers; GPU access and audio-device access still require host configuration.
1 introductory training example · not runtime-testedExplore with Keeks ↗Azure
Deploy cloud workloads with explicit resources and credentials; select the service and region before estimating capabilities or cost.
1 introductory training example · not runtime-testedExplore with Keeks ↗OpenAI
Treat this as a provider and API ecosystem; ask for the exact model and endpoint, and keep API keys server-side.
1 introductory training example · not runtime-testedExplore with Keeks ↗.NET
Build applications with the .NET runtime and libraries; distinguish the runtime from C#, one language targeting it.
1 introductory training example · not runtime-testedExplore with Keeks ↗Kusto Query Language
Query telemetry and time-series logs; check time windows and aggregation grain before interpreting latency percentiles.
1 introductory training example · not runtime-testedExplore with Keeks ↗gRPC
Define typed remote services, often using Protocol Buffers; account for deadlines, streaming behavior and versioned schemas.
1 introductory training example · not runtime-testedExplore with Keeks ↗REST APIs
Design resource-oriented HTTP interfaces; validate inputs and use explicit status codes and authentication boundaries.
1 introductory training example · not runtime-testedExplore with Keeks ↗OBS
Compose and stream audiovisual scenes; inspect audio routing, sync offsets and encoder load separately.
1 introductory training example · not runtime-testedExplore with Keeks ↗NDI
Transport video and related media over a network; bandwidth, network conditions and buffering affect end-to-end latency.
1 introductory training example · not runtime-testedExplore with Keeks ↗OBS Camera
Clarify whether this means OBS Virtual Camera or a specific phone-camera app; the name alone is insufficient for reliable setup instructions.
1 introductory training example · not runtime-testedExplore with Keeks ↗Protocols & concepts
MIDI
Represent musical performance and control events; note-on velocity is not an audio waveform and channel routing affects instrument selection.
1 introductory training example · not runtime-testedExplore with Keeks ↗OSC
Send addressed control messages between applications; specify argument types, transport and timing expectations at both endpoints.
1 introductory training example · not runtime-testedExplore with Keeks ↗SDFs (signed distance functions)
Represent distance to a surface with a sign convention; sphere tracing requires a valid distance bound, so arbitrary procedural fields can overshoot.
1 introductory training example · not runtime-testedExplore with Keeks ↗No matching tools. Try a shorter name or a category.
How training becomes a model
Review and expand examples → run the Colab GPU notebook → compare base and adapter answers → deploy the adapter. Adding a tool to this page alone does not train a model.
Open training notebook ↗ · Changelog ↗