publish assertion encoder with model card

README.md

@@ -0,0 +1,146 @@
+---
+language: en
+library_name: setfit
+pipeline_tag: text-classification
+base_model: sentence-transformers/paraphrase-mpnet-base-v2
+datasets:
+  - StanfordSCALE/assertions_llm_annotated_talkmoves
+metrics:
+  - f1
+  - precision
+  - recall
+  - roc_auc
+tags:
+  - setfit
+  - text-classification
+  - education
+  - classroom-discourse
+  - edubehaviors
+  - assertion
+model-index:
+- name: assertion_proposes_a_problem_solving_strategy
+  results:
+  - task:
+      type: text-classification
+    dataset:
+      name: assertions_llm_annotated_talkmoves
+      type: StanfordSCALE/assertions_llm_annotated_talkmoves
+      split: test
+    metrics:
+    - type: f1
+      value: 0.559
+      name: F1 (positive class, test)
+    - type: precision
+      value: 0.595
+      name: Precision (positive class, test)
+    - type: recall
+      value: 0.5272
+      name: Recall (positive class, test)
+    - type: roc_auc
+      value: 0.8768
+      name: ROC-AUC (test)
+
+---
+
+# Assertion: proposes a problem solving strategy
+
+Assertion id: `bu5__proposes_a_problem_solving_strategy` (bottom-up assertion set `bu5`).
+
+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](https://github.com/SumnerLab/TalkMoves). See the Datasets section below for more information.
+
+---
+
+## Training Details
+
+### Datasets
+
+| Dataset | Split | Size |
+|---------|-------|------|
+| [StanfordSCALE/assertions_llm_annotated_talkmoves](https://huggingface.co/datasets/StanfordSCALE/assertions_llm_annotated_talkmoves) | train | 13,180 (53.3%) |
+| [StanfordSCALE/assertions_llm_annotated_talkmoves](https://huggingface.co/datasets/StanfordSCALE/assertions_llm_annotated_talkmoves) | dev | 3,296 (13.3%) |
+| [StanfordSCALE/assertions_llm_annotated_talkmoves](https://huggingface.co/datasets/StanfordSCALE/assertions_llm_annotated_talkmoves) | test | 8,239 (33.3%) |
+
+This model's columns are `assertion_proposes_a_problem_solving_strategy` and `split_proposes_a_problem_solving_strategy`.
+
+**Base rate** (share of rows labeled as True): 11.4% overall —
+11.6% train, 10.7% dev, 11.4% test.
+
+### Labels and annotation
+
+Labels were generated with LLM annotators. Krippendorff's alpha for this assertion is 0.569.
+
+
+### Hyperparameters
+
+| Parameter | Value |
+|-----------|-------|
+| Base model (body) | [sentence-transformers/paraphrase-mpnet-base-v2](https://huggingface.co/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 | 3,296 | 10.7% | 0.570 | 0.425 | 0.487 | 0.861 | 0.480 |
+| test | 8,239 | 11.4% | 0.595 | 0.527 | 0.559 | 0.877 | 0.564 |
+
+### Limitations
+
+- **Labels come from LLM annotators, not human coders.** Agreement between annotators with Krippendorff's Alpha is 0.569.
+- 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
+
+```bash
+pip install setfit
+```
+
+```python
+from setfit import SetFitModel
+
+model = SetFitModel.from_pretrained("StanfordSCALE/assertion_proposes_a_problem_solving_strategy")
+
+text = 'But one thing that would be  the same is the relative one thing you have to be careful each of you  in your own models would be the sizes in relation to the other sizes  right'
+model.predict([text])        # -> array([1]) when the assertion holds
+model.predict_proba([text])  # -> [[P(no), P(yes)]]
+```
+
+
+## Citation
+
+```bibtex
+@misc{assertion_proposes_a_problem_solving_strategy,
+  author = {Stanford SCALE Initiative},
+  title  = {Assertion classifier: proposes a problem solving strategy},
+  year   = {2026},
+  url    = {https://huggingface.co/StanfordSCALE/assertion_proposes_a_problem_solving_strategy}
+}
+```