LLM review¶

LLMReview¶

class pysyrev.LLMReview(text_inputs, reviewers, workflow_schema, export, decision_rule, batch_size, api_pause, sample_size, doc_dataset)[source]¶

Bases: object

Runtime LLM review pipeline.

Attributes mirror what ReviewConfig describes, plus the few derived runtime objects (reviewers, workflow_schema). Use from_config to build it from a parsed ReviewConfig — direct construction is also supported for tests or programmatic use.

Parameters:

text_inputs (List[str])
reviewers (List[Reviewer])
workflow_schema (List[dict])
export (ReviewExportConfig)
decision_rule (str)
batch_size (int)
api_pause (float)
sample_size (int | None)
doc_dataset (str | None)

text_inputs: List[str]¶

reviewers: List[Reviewer]¶

workflow_schema: List[dict]¶

export: ReviewExportConfig¶

decision_rule: str¶

batch_size: int¶

api_pause: float¶

sample_size: int | None¶

doc_dataset: str | None¶

classmethod from_config(config)[source]¶

Build an LLMReview from a parsed ReviewConfig.

Return type:: LLMReview
Parameters:: config (ReviewConfig)

run(dataset=None)[source]¶: Execute the review. If dataset is None, load from doc_dataset.

save()[source]¶

property included_docs: DataFrame¶

ReviewedDataset¶

class pysyrev.ReviewedDataset(total_docs=None, included_docs=None)[source]¶

Bases: object

Parameters:

total_docs (DataFrame)
included_docs (DataFrame)

total_docs: DataFrame = None¶

included_docs: DataFrame = None¶