### gemma2-9b · medial · geo
   idiom config:    {'model': 'google/gemma-2-9b', 'reduction': 'geometric_mean', 'medial_only': True, 'dtype': 'bfloat16', 'dataset': '/home/prada/PID_evaluation/data/dataset.tsv', 'num_idioms': 18, 'syn_reg_eps': 0.01}
   nonidiom config: {'model': 'google/gemma-2-9b', 'reduction': 'geometric_mean', 'medial_only': True, 'dtype': 'bfloat16', 'dataset': '/home/prada/PID_evaluation/data/nonidioms_dataset.tsv', 'num_idioms': 18, 'syn_reg_eps': 0.01}

== idioms :: ratio_u_idiom ==  (N=18 phrases)
     mean     median                    95% CI
   1.2395     1.2340  [  1.2031,  1.2763]

== non-idioms :: ratio_u_idiom ==  (N=18 phrases)
     mean     median                    95% CI
   1.0911     1.0981  [  1.0679,  1.1153]
  cross-dataset ratio_u_idiom: idioms - nonidioms  Δ=+0.1484  CI=[+0.1057,+0.1925]  *

== idioms :: ratio_s_idiom ==  (N=13 phrases)  (5 non-finite dropped)
     mean     median                    95% CI
   1.1242     1.1086  [  1.1003,  1.1501]

== non-idioms :: ratio_s_idiom ==  (N=3 phrases)  (15 non-finite dropped)
     mean     median                    95% CI
   1.3018     1.2479  [  1.1131,  1.5443]
  cross-dataset ratio_s_idiom: idioms - nonidioms  Δ=-0.1776  CI=[-0.4136,+0.0058]