🚀 Generate chat completions and analyze results.
⚙️ Customize parameters such as temperature and top-p.
⏳ Record and label execution histories for easy comparison.
🆚 Conduct batch testing of prompts for streamlined optimization.
🧩 Incorporate variables for enhanced flexibility and precision in prompt crafting.
𝒇 Mimic external function calls for comprehensive testing scenarios.
Features
Some key features and benefits of Prompter:
Lightweight & User-Friendly
Prompter is lightweight and runs fully in the browser without installation. It also boasts a stylish and intuitive UI/UX design that makes prompt engineering fast and easy.
Customizable Parameters
Fine-tune parameters like temperature, top-p, etc. to shape model responses.
Prompt History
Save and restore past prompts for reference. Easily re-run previous tests.
Variables & Function Calling
Adjust prompts dynamically and simulate the interaction with external APIs. This also includes support for OpenAI's feature of making parallel function calling.
Batch Testing
Test multiple prompts in batch mode for faster iteration.
Custom Endpoints
Use your own API endpoint in addition to the official OpenAI API.
Seed & JSON format
Prompter supports seed and system fingerprint, and offers JSON output format for enhanced functionality and data handling.
🌟 We believe developers deserve the best – a tool that's both efficient and aesthetically pleasing.
Tutorial
How to use Prompter
Manual
1. Setup API Key
At Prompter, we prioritize your security.
Your API key is safely stored and never accessed, ensuring your sensitive information remains private and protected.
2. Create Prompt
The prompt editor has four main sections👇
Parameters
Configurations like model, max tokens, temperature that shape the API request.
Variables
Create a new variable by entering the variable name and its value. When using the variables, enclose the variable name with double curly braces {{ and }} in the Messages tab. You can also test variables in batches by clicking the Batch Request button.
Functions
In order to simulate Function Calling, please create a function by entering the function name and description. And then input the JSON schema of function definition and the mock value accordingly.
3. Tweak and Iterate
To start, accept the default parameters or tweak them as needed. Then enter your prompts and hit "Run" to submit the requests. The result will appear on the right, along with a copy button.
Now you can iterate on your prompt:
Change parameters like Temperature and Top-p.
Edit the prompts and resend.
Restore past prompts from history for comparison.
4. Results & History
You can easily label the results as "like" or "dislike". In the History, you can quickly filter based on these labels, making it more convenient to select the most appropriate parameters and prompts by comparing the outputs under different input scenarios.
5. Batch Testing ✨
To test multiple prompts at once, you can click the batch request button to switch to Batch Mode. This lets you test many variations efficiently in parallel. Results will display sequentially as they are returned. You can also copy any individual result.
6. Prompt Management
The prompts can be organized and managed into different projects/groups in the left side bar as below.
Misc
Changelog
v
1.0.0
Nov. 28th 2023
Officially launched 🚀
v
0.0.3
Oct. 11th 2023
Support function calling
v
0.0.2
Sep. 4th 2023
Variables & labelling
Support variables;
Results labelling and filter;
Minor improvements;
v
0.0.1
Oct. 11th 2023
Alpha release