Agents

Guide about Agents to be used in Cybertron and best practices for using them

How to use

• Go to New Assistant Config
• Name your assistant
• Select an agent from the Agent Type list

  

• Press Save

Comparison of Agents

Claude 2 => anthropic.claude-v2

• Use cases like: thoughtful dialogue, content creation, complex reasoning, creativity, and coding.
• Enhanced coding, math, and reasoning skills
• Capable of handling up to 100K tokens in each prompt
• Offers a friendly, enthusiastic conversational experience

Gemini 1 Pro => Gemini 1.0 Pro

• Can handle multi-turn conversational format.
• Can handle zero, one, and few-shot tasks.
• input/output is: text and generates text
• input/output token limit: 30720/2048
• Model last updated: February 2024

GPT 3.5 Turbo => gpt-3.5-turbo-0125

• Optimized for chat/dialogue
• Improved accuracy in responding in requested formats
• Bug fixes for non-English language function calls
• Context window up to 16,385 tokens
• Training data up to Sep 2021

GPT 4 (Stable) => gpt-4-0613

• Large multimodal model
• Solves difficult problems with greater accuracy
• Optimized for chat and traditional completions tasks
• Context window up to 8,192 tokens
• Training data up to Sep 2021

GPT 4 0125 => gpt-4-0125-preview

• intended to reduce cases of “laziness” where the model doesn’t complete a task
• More complete task completions, such as code generation
• Returns a maximum of 4,096 output tokens
• Context window up to 128,000 tokens
• Training data up to Dec 2023

GPT 4 1106 => gpt-4-1106-preview

• Used for longer conversations or documents
• featuring improved instruction following, JSON mode, reproducible outputs, parallel function calling, and more.
• Returns a maximum of 4,096 output tokens.
• Context window up to 128,000 tokens
• Training data up to Apr 2023

GPT 4 Turbo => gpt-4-turbo-2024-04-09

• The latest GPT-4 Turbo model with vision capabilities
• Vision requests can now use JSON mode and function calling
• Context window up to 128,000 tokens
• Training data up to Dec 2023