This dataset contains reward model analysis results for IRL training.
Base Model ID: ajagota71/toxicity-reward-model-v-head-output-max-margin-seed-42-pythia-70m
Full Model ID: ajagota71/toxicity-reward-model-v-head-output-max-margin-seed-42-pythia-70m
Epoch: 0
Analysis Timestamp: 2025-08-03T13:03:24.996419
Number of Samples: 18000
sample_index: Index of the sample
prompt: Input prompt (if available)… See the full description on the dataset page:
https://huggingface.co/datasets/ajagota71/pythia-70m-epochs-0-39-p3-O.
