v0.1.0

contextual-distillation

Extract structured, query-relevant context from documents using LLM-backed structured output.

example.py

from contextual_distillation import Distiller
from pydantic import BaseModel

class Answer(BaseModel):
    summary: str
    key_points: list[str]

distiller = Distiller(
    model_name="openai/gpt-4.1-mini",
    system_prompt="Extract relevant information.",
    response_schema=Answer,
)

result = await distiller.distill(
    query="What are the requirements?",
    chunk_text="Students must complete 130 credits...",
)
print(result.summary)

Structured Extraction

Extract structured data from unstructured documents using LLM-backed Pydantic schemas.

Progressive Refinement

Funnel approach: start broad, narrow at each stage with prompt-guided filtering.

Chunk Processing

Split documents into chunks, process in parallel with WorkerPool, aggregate results.

Observable

Built-in metrics collection, SQLite storage, and web dashboard for monitoring.