EvoGrad: An Online Platform for an Evolving Winograd Schema Challenge using Adversarial Human Perturbations
Bibliographic record
Abstract
Transformer-based language models have been recently showcasing impressive performance on a number of common-sense reasoning tasks such as the Winograd Schema Challenge (WSC) while continuing to struggle on task instances that are either slightly re-worded or perturbed. In the following paper, we wish to address these issues by re-framing the WSC using a never-ending learning, human-in-the-loop scenario devised specifically for perturbed pronoun co-reference resolution problems. We introduce EvoGrad, an open-source, user-friendly platform for the continual evaluation and development of models, based on iterations of human-adversarial perturbations. Given that common-sense knowledge varies cross-culturally and through time, our platform allows for the communal contribution towards an evolving task that is both inclusive and accessible to wider societies. In addition, we propose a novel mechanism to develop new task instances, and define a new metric to measure model stability on such dynamic tasks, called the Minimum Error Depth. We show that models fine-tuned on a small iteration of EvoGrad have their performance boosted on WSC-based tasks; this indicates a promising synergy between the acquisition of common sense and the never-ending learning paradigm.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.007 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".