CoolFace
Modelpublic

Pritish92/assignment2-artifacts

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face
0likes
Model Card

Assignment 2 Artifacts

Experiment artifacts for Safety Alignment in LLMs.

Contents

FileDescription
function_vector.ptFinal Function Vector for activation steering
aie_scores.ptAIE scores for all (layer, head) pairs
mean_clean.ptMean clean projected head contributions
aie_heatmap.pngAIE heatmap visualization
part1_*.jsonSFT and DARE training metadata
part2_*.jsonHarmful model and RESTA metadata
part3_*.jsonFunction Vector extraction metadata
part4_*.jsonEvaluation results (safety + utility)
Part_*.ipynbFinal executed notebooks for the four assignment parts
Report.pdfFinal report PDF
22MF3IM15_Assignment_2.zipFinal submission zip

Student: 22MF3IM15