llmon -- lang/unstructured_text
Copyright 2026 IBM Corp.

This product includes software developed as part of the LLMON project
(https://github.com/generative-computing/llmon), licensed under the Apache
License, Version 2.0. See LICENSE.

--------------------------------------------------------------------------------
Third-party models
--------------------------------------------------------------------------------

The `tagger` backend fine-tunes ModernBERT-base
(https://huggingface.co/answerdotai/ModernBERT-base), distributed by Answer.AI
and LightOn under the Apache License, Version 2.0. Checkpoints published for this
project are derivative works of that model and carry the same license.

--------------------------------------------------------------------------------
Training data
--------------------------------------------------------------------------------

No training or evaluation corpus is distributed with this project. The published
tagger checkpoint was trained on span annotations produced for this project,
layered over prompts drawn from four upstream benchmarks, and docs/guidelines.html
quotes example prompts from them. That prompt text remains subject to its
original terms:

  IFEval                    Apache License, Version 2.0
    Zhou et al., "Instruction-Following Evaluation for Large Language Models",
    arXiv:2311.07911.  https://huggingface.co/datasets/google/IFEval

  MMLU                      MIT License
    Hendrycks et al., "Measuring Massive Multitask Language Understanding",
    arXiv:2009.03300.  https://huggingface.co/datasets/cais/mmlu

  InfoBench                 see upstream
    Qin et al., "InFoBench: Evaluating Instruction Following Ability in Large
    Language Models", arXiv:2401.03601.

  ComplexBench              see upstream
    Wen et al., "Benchmarking Complex Instruction-Following with Multiple
    Constraints Composition", arXiv:2407.03978.

The two marked "see upstream" are recorded here without a license assertion
because their terms were not verified by the authors of this NOTICE. Consult each
project before reusing that text in a context those terms do not permit.
