Counterfactual Faithfulness on SNLI
0.243Faithfulness ScoreTulu-2 13B
Evaluation results
| Method | Links | |
|---|---|---|
| Tulu-2 13BTraining Setup=No-Training, Word Constraint=One-word2025.12 | 0.243 | |
| Tulu-2 13BTraining Setup=w/ Predictions, Word Constraint=One-word2025.12 | 0.216 | |
| Tulu-2 13BTraining Setup=w/ Explanations, Word Constraint=One-word2025.12 | 0.192 | |
| Tulu-2 7BTraining Setup=w/ Explanations, Word Constraint=One-word2025.12 | 0.17 | |
| Tulu-2 7BTraining Setup=w/ Predictions, Word Constraint=One-word2025.12 | 0.079 | |
| Tulu-2 7BTraining Setup=No-Training, Word Constraint=One-word2025.12 | 0.076 |