Output Tracking Table

This table maps each test image to its model response over multiple prompt versions.

Image ID Prompt Version Output (Summary) Variance Notes
img_01 v6 R1: l=0.54, r=0.60, R2: l=… Minor float noise
img_01 v9 R1: l=0.56, r=0.61, R2: l=… Slight change in R2, R6
img_02 v6 R1: l=0.35, r=0.40, R2: l=… Consistent across versions
img_02 v9 R1: l=0.35, r=0.40, R2: l=… Deterministic

Notes

  • Use image filenames or UUIDs for reproducibility
  • Helps identify prompt regression or version drift