### gpt2 · medial · geo
   idiom config:    {'model': 'gpt2', 'reduction': 'geometric_mean', 'medial_only': True, 'dtype': 'float32', 'dataset': '/home/prada/PID_evaluation/data/dataset.tsv', 'num_idioms': 18, 'syn_reg_eps': 0.01}
   nonidiom config: {'model': 'gpt2', 'reduction': 'geometric_mean', 'medial_only': True, 'dtype': 'float32', '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
   1.1425     1.1397  [  1.1101,  1.1789]

== non-idioms :: ratio_u_idiom ==  (N=18 phrases)
     mean     median                    95% CI
   1.0590     1.0579  [  1.0454,  1.0733]
  cross-dataset ratio_u_idiom: idioms - nonidioms  Δ=+0.0835  CI=[+0.0481,+0.1223]  *

== idioms :: ratio_s_idiom ==  (N=15 phrases)  (3 non-finite dropped)
     mean     median                    95% CI
   1.2110     1.1894  [  1.1769,  1.2446]

== non-idioms :: ratio_s_idiom ==  (N=8 phrases)  (10 non-finite dropped)
     mean     median                    95% CI
   1.3518     1.3581  [  1.2884,  1.4155]
  cross-dataset ratio_s_idiom: idioms - nonidioms  Δ=-0.1408  CI=[-0.2135,-0.0686]  *