← Back to Tools & Services
ToolActive

RT-SEG

A toolkit for reasoning trace segmentation

Maintained by Leon Hammerla

Developed by Leon Hammerla, Bhuvanesh Verma, Alexander Mehler

Software · 2026

About this tool

RT-SEG is a modular toolkit for segmenting reasoning traces into coherent units and optionally assigning labels to the resulting segments. It provides rule-based, probabilistic, semantic, and LLM-based segmentation engines together with strategies for combining segment boundaries from multiple engines.

Tags

reasoning tracessegmentationLLMchain of thoughtNLPdiscourse segmentationneglab

Related entries

BibTeX

@project{INF:RT-SEG,
  keywords      = {tool},
  name          = {RT-SEG},
  subtitle      = {A toolkit for reasoning trace segmentation},
  author        = {Hammerla, Leon and Verma, Bhuvanesh and Mehler, Alexander},
  maintainer    = {Hammerla, Leon},
  year          = {2026},
  abstract      = {RT-SEG is a modular toolkit for segmenting reasoning traces
                   into coherent units and optionally assigning labels to the
                   resulting segments. It provides rule-based, probabilistic,
                   semantic, and LLM-based segmentation engines together with
                   strategies for combining segment boundaries from multiple
                   engines.},
  tooltype      = {software},
  url           = {https://texttechnologylab.github.io/RT-SEG/},
  documentation = {https://texttechnologylab.github.io/RT-SEG/},
  projectlink   = {https://github.com/texttechnologylab/RT-SEG},
  access        = {Public},
  tags          = {reasoning traces, segmentation, LLM, chain of thought,
                   NLP, discourse segmentation, neglab},
  status        = {active},
  license       = {MIT},
  related       = {Hammerla:etal:2026:c}
}