Web Playground
The Trunkate AI Playground is your interactive environment for testing prompt optimizations before deploying them in code.
Accessing the Playground
Navigate to http://trunkate.ai/#demo to access the interactive playground.
Key Features
Live Optimization
Type or paste your raw, verbose prompt into the input area. The playground will immediately send it through the Trunkate pipeline and display the optimized version side-by-side.
Token Savings Metrics
Below the text areas, you can see real-time statistics:
- Original Tokens: The size of your initial prompt.
- Optimized Tokens: The size after Trunkate processes it.
- Reduction %: The total percentage of tokens saved.
Adjusting the Budget
In the settings panel, you can simulate API behavior by adjusting the Target Budget.
- Strict (Low Budget): Aggressive pruning and structuring. Best for GPT-4 where costs are high.
- Relaxed (High Budget): Only removes the most obvious fluff. Best when preserving maximum linguistic nuance is preferred.
Last updated on