task stringclasses 3
values | input stringclasses 3
values | target stringclasses 3
values |
|---|---|---|
uncertainty | What shirt suits blue eyes? | Ask for skin tone, hair color before answering. |
context | Project state? | Use latest verified state, not older context. |
math | 7x9 | 63 |
Parameter Golf Auxiliary Dataset (PG-CCE)
This dataset is designed to reduce entropy in language models by addressing three critical failure modes:
- Hallucination / uncertainty calibration
- Long-context consistency and state tracking
- Exact discrete reasoning (numbers, code, structured patterns)
It is intended as a small auxiliary training dataset to improve bits-per-byte (BPB) performance in Parameter Golf.
Usage
Mix as auxiliary data:
- 97% main dataset (FineWeb)
- 3% PG-CCE
Do not replace the main dataset.
Structure
Each record contains:
- task
- input
- target
Goal
Reduce probability mass on incorrect but plausible outputs and improve structural precision.
- Downloads last month
- 38