StanfordSCALE/assertion_agrees_with_one_or_more_students_answer

Assertion: agrees with one or more students answer

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

Dataset	Split	Size
StanfordSCALE/assertions_llm_annotated_talkmoves	train	13,185 (53.3%)
StanfordSCALE/assertions_llm_annotated_talkmoves	dev	3,291 (13.3%)
StanfordSCALE/assertions_llm_annotated_talkmoves	test	8,239 (33.3%)

This model's columns are assertion_agrees_with_one_or_more_students_answer and split_agrees_with_one_or_more_students_answer.

Base rate (share of rows labeled as True): 7.1% overall — 6.6% train, 8.5% dev, 7.4% test.

Labels and annotation

Labels were generated with LLM annotators. Krippendorff's alpha for this assertion is 0.745.

Hyperparameters

Parameter	Value
Base model (body)	sentence-transformers/paraphrase-mpnet-base-v2
Head	LogisticRegression
Body learning rate	2e-05
Head learning rate	0.01
Batch size	16 (contrastive phase) / 32 (head)
Epochs	10
Max steps	5000 (contrastive phase)
Eval max steps	100
Seed	20260904
Mixed precision	enabled on GPU

---

Evaluation

Results

Split	n	Base rate	Precision	Recall	F1 (positive class)	ROC-AUC	Average precision
dev	3,291	8.5%	0.534	0.366	0.434	0.858	0.437
test	8,239	7.4%	0.572	0.333	0.421	0.852	0.446

Limitations

• Labels come from LLM annotators, not human coders. Agreement between annotators with Krippendorff's Alpha is 0.745.
• Trained on teacher utterances only. Behaviour on student speech is untested.

---

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_agrees_with_one_or_more_students_answer")

text = 'Yeah can you get different answers right Michael and Erik'
model.predict([text])        # -> array([1]) when the assertion holds
model.predict_proba([text])  # -> [[P(no), P(yes)]]

Citation

@misc{assertion_agrees_with_one_or_more_students_answer,
  author = {Stanford SCALE Initiative},
  title  = {Assertion classifier: agrees with one or more students answer},
  year   = {2026},
  url    = {https://huggingface.co/StanfordSCALE/assertion_agrees_with_one_or_more_students_answer}
}