Back to Experiments

Create New Experiment

Configure your LLM experiment parameters and run it to generate responses with quality metrics.

Experiment Details
Provide the basic information and prompt for your LLM experiment

A descriptive name for your experiment

Optional description of your experiment goals

The prompt that will be sent to the LLM

Parameter Ranges
Configure the parameter ranges for your experiment. The system will automatically generate combinations within these ranges and test each one to find the optimal settings.

Temperature

Controls randomness

Controls the randomness of the output. Lower values (0.1-0.3) produce more focused, deterministic responses. Higher values (0.7-1.0) produce more creative, diverse responses.

Top-P (Nucleus Sampling)

Controls diversity

Controls the diversity of token selection. Lower values (0.1-0.3) focus on high-probability tokens, while higher values (0.7-0.9) allow more diverse token selection.

Top-K

Controls vocabulary

Limits the number of highest-probability tokens to consider. Lower values (10-20) focus on the most likely tokens, while higher values (40-50) allow more vocabulary diversity.

Max Tokens

Controls length

Sets the maximum length of the generated response. Lower values (10-50) produce concise answers, while higher values (100-500) allow for more detailed, comprehensive responses.