bio_reasoning.reasoning package#

Reasoning Mode defines 1) the available tools and 2) the system prompt for the LLM in this specific mode. System prompt could contain instruction on which order to use the tools, or which tool to use first, etc. Tools are presented as a ToolRegistry instance, which will handle the tool selection and execution.

class bio_reasoning.reasoning.ReasoningMode(*, layer_a: ToolRegistry, layer_b: ToolRegistry, layer_c: ToolRegistry, sys_prompt: str, name: str = 'Generic Reasoning Mode')[source]#

Bases: object

Base class that encapsulate general reasoning mode elements and methods

__init__(*, layer_a: ToolRegistry, layer_b: ToolRegistry, layer_c: ToolRegistry, sys_prompt: str, name: str = 'Generic Reasoning Mode')[source]#
property layers: ToolRegistry#

Present the merged layers as a single ToolRegistry instance. This allows the user to access all the tools in the reasoning mode.

class bio_reasoning.reasoning.ExampleReasoningMode[source]#

Bases: ReasoningMode

An example of a specific reasoning mode that might use certain tools. Show case the process of building a reasoning mode. We essentially need to define the tools for each layers, and the system prompt.

__init__()[source]#

Submodules#