rscl
Datasets
All datasets matching “rscl”rscl-error-recognition-fidelity-v0.1
What this dataset tests
Whether a system recognizes its own error when challenged.
Not whether it can produce a new answer.
Recognition first.
Why this exists
Self-correction fails early when the model:
denies the mistake
misses the actual error type
changes output without acknowledging why
This dataset scores the recognition step.
Data format
Each row contains:
user_task
model_initial_output
error_signal
model_acknowledgement
error_type… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/rscl-error-recognition-fidelity-v0.1.rscl-correction-quality-under-constraint-v0.1
What this dataset tests
Whether a correction is both:
correct
compliant with constraints
A model can fix the contentand still violate the rules.
This dataset separates those cases.
Why this exists
Self-correction often fails by:
fixing the answer but breaking the format
complying with the format but keeping the error
introducing new errors during repair
This benchmark scores correction quality under constraint.
Data format
Each row contains:… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/rscl-correction-quality-under-constraint-v0.1.