### llama3.1-8b · medial · geo
   idiom config:    {'model': 'meta-llama/Llama-3.1-8B', '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': 'meta-llama/Llama-3.1-8B', '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
   1.2072     1.1966  [  1.1794,  1.2361]

== non-idioms :: ratio_u_idiom ==  (N=18 phrases)
     mean     median                    95% CI
   1.0703     1.0648  [  1.0529,  1.0890]
  cross-dataset ratio_u_idiom: idioms - nonidioms  Δ=+0.1369  CI=[+0.1038,+0.1709]  *

== idioms :: ratio_s_idiom ==  (N=17 phrases)  (1 non-finite dropped)
     mean     median                    95% CI
   1.1701     1.1690  [  1.1498,  1.1905]

== non-idioms :: ratio_s_idiom ==  (N=5 phrases)  (13 non-finite dropped)
     mean     median                    95% CI
   1.3164     1.2803  [  1.2085,  1.4508]
  cross-dataset ratio_s_idiom: idioms - nonidioms  Δ=-0.1463  CI=[-0.2831,-0.0386]  *