agents.reasoning_and_critique.tot.models¶

Tree of Thoughts (ToT) models and data structures.

This module defines the core data models for the Tree of Thoughts reasoning algorithm, including candidate solutions and structured output models.

Classes¶

`Candidate`	A candidate solution in the Tree of Thoughts algorithm.
`CandidateEvaluation`	Structured output model for evaluating a candidate solution.
`CandidateGeneration`	Structured output model for generating multiple candidate solutions.
`Equation`	An equation in reverse-polish notation that combines numbers to reach a target value.
`EquationGeneration`	Specialized candidate generation for equation problems.
`Score`	A score with feedback.
`ScoredCandidate`	A candidate solution with its evaluation score.

Functions¶

update_candidates([existing, updates])

Update candidate list, handling special cases like clearing.

Module Contents¶

class agents.reasoning_and_critique.tot.models.Candidate(/, **data)¶

Bases: pydantic.BaseModel, Generic[T]

A candidate solution in the Tree of Thoughts algorithm.

This represents an unscored candidate solution generated by the LLM.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class agents.reasoning_and_critique.tot.models.CandidateEvaluation(/, **data)¶

Bases: pydantic.BaseModel

Structured output model for evaluating a candidate solution.

This model is used when the evaluator LLM produces structured output.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

to_score()¶

Convert to a Score object.

Return type:: Score

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class agents.reasoning_and_critique.tot.models.CandidateGeneration(/, **data)¶

Bases: pydantic.BaseModel, Generic[T]

Structured output model for generating multiple candidate solutions.

This model is used when the generator LLM produces structured output.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

to_candidates()¶

Convert candidate contents to Candidate objects.

Return type:: list[Candidate[T]]

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class agents.reasoning_and_critique.tot.models.Equation(/, **data)¶

Bases: pydantic.BaseModel

An equation in reverse-polish notation that combines numbers to reach a target value.

This is one possible implementation of a solution type for mathematical problems.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

compute()¶

Compute the result of the equation.

Return type:: float

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class agents.reasoning_and_critique.tot.models.EquationGeneration(/, **data)¶

Bases: pydantic.BaseModel

Specialized candidate generation for equation problems.

This is a structured output model for equation-specific generation.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

to_candidates()¶

Convert equations to Candidate objects.

Return type:: list[Candidate[Equation]]

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class agents.reasoning_and_critique.tot.models.Score(/, **data)¶

Bases: pydantic.BaseModel

A score with feedback.

This represents the evaluation of a candidate solution.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

class agents.reasoning_and_critique.tot.models.ScoredCandidate(/, **data)¶

Bases: pydantic.BaseModel, Generic[T]

A candidate solution with its evaluation score.

This combines a candidate solution with its score and feedback.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

property content: T¶

Get the content of the candidate.

Return type:: T

property feedback: str | None¶

Get the feedback if any.

Return type:: str | None

property metadata: dict[str, Any]¶

Get the metadata of the candidate.

Return type:: dict[str, Any]

model_config¶: Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

property value: float¶

Get the numerical score value.

Return type:: float

agents.reasoning_and_critique.tot.models.update_candidates(existing=None, updates=None)¶

Update candidate list, handling special cases like clearing.

Parameters:

existing (list[Any] | None) – Current list of candidates
updates (list[Any] | str | None) – New candidates to add, or “clear” to empty the list

Returns:

Updated list of candidates

Return type:

list[Any]