​
SelfInstructGenerator

class SelfInstructGenerator(BaseGenerator):
A generator for creating synthetic datapoints using self-instruct. It utilizes both a human-provided dataset (seed_dataset) and generated machine instructions (machine_instructions) to produce new, synthetic datapoints that include a question, a computed rationale (code), and a final answer (from a verifier).

​
init

def __init__(
    self,
    seed_dataset: StaticDataset,
    verifier: BaseVerifier,
    instruction_agent: Optional[ChatAgent] = None,
    rationale_agent: Optional[ChatAgent] = None,
    seed: int = 42,
    **kwargs
):
Initialize the self-instruct generator. Parameters:
  • seed_dataset (StaticDataset): Dataset containing seed instructions.
  • verifier (BaseVerifier): Verifier instance to validate generated solutions.
  • instruction_agent (Optional[ChatAgent]): Agent for generating instructions. If not provided, a default agent will be created.
  • rationale_agent (Optional[ChatAgent]): Agent for generating rationales. If not provided, a default agent will be created.
  • seed (int): Random seed for reproducibility. (default: :obj:42) **kwargs: Additional keyword arguments passed to the BaseGenerator. (default: 42)

​
default_instruction_agent

def default_instruction_agent(self):
Returns: ChatAgent: An agent with the default instruction prompt.

​
default_rationale_agent

def default_rationale_agent(self):
Returns: ChatAgent: An agent with the rationale prompt

​
format_support_block

def format_support_block(dp: DataPoint):
Format a DataPoint into a few-shot example block. Parameters:
  • dp (DataPoint): A data point.
Returns: str: A formatted string containing the question and its corresponding code block in Markdown-style Python format.

​
generate_new_instruction

def generate_new_instruction(
    self,
    agent: ChatAgent,
    support_human_dps: list[DataPoint],
    support_machine_dps: list[DataPoint]
):
Generate a new instruction using self-instruct prompting. Parameters:
  • agent (ChatAgent): The agent to use for generating the instruction.
  • support_human_dps (list[DataPoint]): List of human examples to sample.
  • support_machine_dps (list[DataPoint]): List of machine examples to sample.
Returns: str: The newly generated question.

​
generate_rationale

def generate_rationale(
    self,
    question: str,
    agent: Optional[ChatAgent] = None,
    support_human_dps: Optional[list[DataPoint]] = None
):
Generate rationale code (solution) for the given question. Parameters:
  • question (str): The question to be solved.
  • agent (Optional[ChatAgent]): The agent to use for generating the rationale. If None is provided, the default rationale agent will be used. (default: :obj:None)
  • support_human_dps (Optional[list[DataPoint]]): List of human examples to sample. (default: :obj:None)
Returns: str: The generated code solution as a string.

​
QuestionSchema

class QuestionSchema(BaseModel):
Schema for the generated question. Parameters:
  • question (str): The question generated by the model.

​
RationaleSchema

class RationaleSchema(BaseModel):
Schema for the generated rationale code. Parameters:
  • code (str): The generated code without any formatting.