neuralset.events.transforms.text.AddSummary¶

class neuralset.events.transforms.text.AddSummary(*, infra: Backend | None = None, model_name: str = 'meta-llama/Llama-3.2-3B-Instruct', n_sentences_requested: int = 3)[source][source]¶

Generate concise summaries for Text events using a pretrained language model.

This transform processes events of type “Text” and generates a summary for each using a large language model (LLM). Summaries are added as new events of type “Summary” in the DataFrame. The summarization prompt ensures that the model returns exactly the requested number of sentences with a precise description of the content.

Parameters:

model_name (str, default="meta-llama/Llama-3.2-3B-Instruct") – Hugging Face model identifier used for text summarization.
n_sentences_requested (int, default=3) – Number of sentences to include in each summary.