StanfordSCALE/assertion_sentence_summarizes_or_reviews

Assertion: sentence summarizes or reviews

This classifier was trained for EduBehaviors: Assertion-based schemas for auditable dialogue coding and is usable through the Python package EduBehaviors-kit. This classifier was trained on an LLM-annotated subset of teacher utterances from the TalkMoves Dataset. See the Datasets section below for more information.

Training Details

Datasets

This model's columns are assertion_sentence_summarizes_or_reviews and split_sentence_summarizes_or_reviews.

Base rate (share of rows labeled as True): 3.2% overall — 3.5% train, 3.0% dev, 2.8% test.

Labels and annotation

Labels were generated with LLM annotators. Krippendorff's alpha for this assertion is 0.229.

Hyperparameters

Evaluation

Results

Limitations

• —Labels come from LLM annotators, not human coders. Agreement with Krippendorff's Alpha is 0.229; this is poor.This model's predictions and the underlying data are unreliable.
• —Trained on teacher utterances only. Behaviour on student speech is untested.
• —Test F1 is 0.128. This model does not work well enough to be used on its own.

How to Use

Message Structure

The model was trained on text built as:

{utterance}

The utterance is passed through as-is.

Running instructions

pip install setfit

from setfit import SetFitModel

model = SetFitModel.from_pretrained("StanfordSCALE/assertion_sentence_summarizes_or_reviews")

text = 'A lot of you guys had really good ideas about how the two diameters related to each other and so you were kind of approaching that'
model.predict([text])        # -> array([1]) when the assertion holds
model.predict_proba([text])  # -> [[P(no), P(yes)]]

Citation

@misc{assertion_sentence_summarizes_or_reviews,
  author = {Stanford SCALE Initiative},
  title  = {Assertion classifier: sentence summarizes or reviews},
  year   = {2026},
  url    = {https://huggingface.co/StanfordSCALE/assertion_sentence_summarizes_or_reviews}
}