Model Selection
Get confidence scores per token when this option is on (requires a Together-backed model).
System Context
Define the system prompt the model will receive
This is the only context the model receives - no hidden instructions
Researcher Settings
How your messages will be labeled in the conversation
Start Experiment
Disconnected
Conversation
Model Metrics
Turns
0
Tokens Used
0
Goal Deviation
0.00
Sentiment Analysis
No sentiment data yet
Start an experiment to see analysis