$ mnemogit --start
Commit to
memory.
AI-powered flashcards with spaced repetition. Choose your model. Compare explanations. Track experiments with MLflow. All from verified sources.
FRONT
Why is positional encoding necessary in transformer architectures?
features
Not another flashcard app.
MnemoGit is a model comparison platform that happens to teach you things. Every explanation is traceable. Every model is testable.
Compare Models Side-by-Side
Run the same card through Claude, GPT, Llama, and Gemma simultaneously. See which model explains best for your learning style.
AI Explanations
Stuck on a concept? Get alternative explanations from any AI model. Filter by capability — reasoning, coding, writing, vision.
SM-2 Spaced Repetition
See cards right when you're about to forget them. Proven to achieve 90%+ long-term retention.
Bring Your Own Key
Use your own API keys for zero markup. Or let us handle it — transparent 3x rates, no hidden fees.
MLflow Experiment Tracking
Every model comparison logs to MLflow. Track which models give you the best results across sessions. Learn MLflow by using it.
Create & Share Decks
Build decks from verified sources or fork from the community library. Every card cites its source.
40 AI/ML Decks
Curated from official docs, open textbooks, and research papers. Foundations through advanced topics.
workflow
Learn like you ship code.
Pick a deck
Choose from 40 curated AI/ML decks or create your own. Every card cites its source.
Compare models
Run the same card through multiple AI models. See which one explains best for you.
Commit to memory
Spaced repetition locks it in. Track your experiments with MLflow.