Model Settings

Model Name: A descriptive name to identify your model
Avatar: Choose a custom avatar icon for your model
System Prompt: Default instructions given to the model

RecurseChat allows you to customize various model settings to optimize performance and output for your specific needs.

Available Settings

Context Length: Maximum number of tokens the model can process at once
Chat Template: Template format used for model inputs
- Controls how user/assistant messages are formatted
- Different models may require specific templates
Temperature: Controls randomness in outputs (0.0-2.0)
- Lower values (0.0-0.3): More focused and deterministic
- Medium values (0.4-0.8): Balanced creativity
- Higher values (0.9-2.0): More random and creative
Top P: Controls diversity of token selection (0.0-1.0)
- Lower values produce more focused outputs
- Higher values allow more diverse responses
Min P: Filters out tokens below this probability threshold (0.0-1.0)
- Higher values limit responses to more likely tokens
- Lower values allow more diverse token selection
Top K: Limits token selection to the K most likely tokens
- Lower values (1-20) produce more focused outputs
- Higher values allow more variety
Max Output Tokens: Maximum number of tokens the model will generate in a single response
Stop Sequences: Custom strings that will cause the model to stop generating

Settings changes take effect immediately without requiring a save button. You can begin using the updated configuration in your next chat session.

Start with conservative context lengths and increase gradually
Match temperature to your use case:
- Use lower values for factual/analytical tasks
- Use higher values for creative writing/brainstorming