Perplexity
Peeklogic AI Orchestrator’s Perplexity integration enables your Salesforce environment to leverage real-time web search capabilities combined with advanced LLMs. Perplexity is ideal for tasks requiring up-to-date information, citations, and detailed research directly within your workflow.
Perplexity Chat Completions
Use this feature to generate responses that require live internet access or specific reasoning capabilities.
Required fields:
- Name: Friendly name for your configuration.
- Model: Select the Perplexity model (e.g., sonar).
Optional fields:
- Reasoning Effort: Controls how much processing power the model deduces to complex reasoning before answering (e.g., low, medium, high).
- search_mode: Defines the scope of the search. The default is web.
- Disable Search: If checked, the model will rely solely on its training data without browsing the live internet.
- Frequency Penalty: Penalizes tokens that have already appeared to reduce repetition.
- Max Tokens: The maximum number of tokens to generate in the response.
- Return Images: If checked, the response may include images relevant to the query.
- Return Related Questions: If checked, the model provides follow-up questions related to the topic.
- Temperature: Controls the randomness of the response.
Search & Date Filters:
These fields allow you to narrow down the search results used by the AI:
- Search Recency Filter: Limits search results to a specific time frame (e.g., month, week, day).
- Search After Date Filter / Search Before Date Filter: restricts search results to a specific date range.
- Last Updated After Filter / Last Updated Before Filter: Filters content based on when it was last updated on the web.
Advanced Sampling Parameters:
- Top K: The number of tokens to keep for top-k filtering. Limits the model to consider only the k most likely next tokens at each step. Lower values (e.g., 10) make the output more focused and deterministic, while higher values allow for more diverse outputs. A value of 0 disables this filter.
- Top P: The nucleus sampling threshold, valued between 0 and 1. Controls the diversity of generated text by considering only the tokens whose cumulative probability exceeds the top_p value. Lower values (e.g., 0.5) make the output more focused, while higher values (e.g., 0.95) allow for more diverse outputs.
