Skip to content

Instantly share code, notes, and snippets.

@mlabonne
Created February 28, 2024 19:22
Show Gist options
  • Save mlabonne/91f68dafff9a89a62a02daec767d2383 to your computer and use it in GitHub Desktop.
Save mlabonne/91f68dafff9a89a62a02daec767d2383 to your computer and use it in GitHub Desktop.
Model AGIEval GPT4All TruthfulQA Bigbench Average
ReversePhiter 35 70.64 48.31 37.97 47.98

AGIEval

Task Version Metric Value Stderr
agieval_aqua_rat 0 acc 20.47 ± 2.54
acc_norm 20.08 ± 2.52
agieval_logiqa_en 0 acc 29.34 ± 1.79
acc_norm 31.18 ± 1.82
agieval_lsat_ar 0 acc 20.87 ± 2.69
acc_norm 22.17 ± 2.75
agieval_lsat_lr 0 acc 34.31 ± 2.10
acc_norm 32.75 ± 2.08
agieval_lsat_rc 0 acc 44.61 ± 3.04
acc_norm 41.26 ± 3.01
agieval_sat_en 0 acc 65.05 ± 3.33
acc_norm 64.56 ± 3.34
agieval_sat_en_without_passage 0 acc 43.20 ± 3.46
acc_norm 39.81 ± 3.42
agieval_sat_math 0 acc 30.45 ± 3.11
acc_norm 28.18 ± 3.04

Average: 35.0%

GPT4All

Task Version Metric Value Stderr
arc_challenge 0 acc 52.13 ± 1.46
acc_norm 53.67 ± 1.46
arc_easy 0 acc 80.64 ± 0.81
acc_norm 77.27 ± 0.86
boolq 1 acc 82.78 ± 0.66
hellaswag 0 acc 56.17 ± 0.50
acc_norm 74.80 ± 0.43
openbookqa 0 acc 37.00 ± 2.16
acc_norm 48.40 ± 2.24
piqa 0 acc 79.87 ± 0.94
acc_norm 80.36 ± 0.93
winogrande 0 acc 77.19 ± 1.18

Average: 70.64%

TruthfulQA

Task Version Metric Value Stderr
truthfulqa_mc 1 mc1 33.41 ± 1.65
mc2 48.31 ± 1.53

Average: 48.31%

Bigbench

Task Version Metric Value Stderr
bigbench_causal_judgement 0 multiple_choice_grade 56.84 ± 3.60
bigbench_date_understanding 0 multiple_choice_grade 57.72 ± 2.58
bigbench_disambiguation_qa 0 multiple_choice_grade 37.60 ± 3.02
bigbench_geometric_shapes 0 multiple_choice_grade 21.73 ± 2.18
exact_str_match 0.00 ± 0.00
bigbench_logical_deduction_five_objects 0 multiple_choice_grade 26.40 ± 1.97
bigbench_logical_deduction_seven_objects 0 multiple_choice_grade 16.43 ± 1.40
bigbench_logical_deduction_three_objects 0 multiple_choice_grade 43.33 ± 2.87
bigbench_movie_recommendation 0 multiple_choice_grade 40.80 ± 2.20
bigbench_navigate 0 multiple_choice_grade 50.80 ± 1.58
bigbench_reasoning_about_colored_objects 0 multiple_choice_grade 56.85 ± 1.11
bigbench_ruin_names 0 multiple_choice_grade 38.17 ± 2.30
bigbench_salient_translation_error_detection 0 multiple_choice_grade 24.95 ± 1.37
bigbench_snarks 0 multiple_choice_grade 69.61 ± 3.43
bigbench_sports_understanding 0 multiple_choice_grade 50.30 ± 1.59
bigbench_temporal_sequences 0 multiple_choice_grade 18.00 ± 1.22
bigbench_tracking_shuffled_objects_five_objects 0 multiple_choice_grade 17.84 ± 1.08
bigbench_tracking_shuffled_objects_seven_objects 0 multiple_choice_grade 12.80 ± 0.80
bigbench_tracking_shuffled_objects_three_objects 0 multiple_choice_grade 43.33 ± 2.87

Average: 37.97%

Average score: 47.98%

Elapsed time: 05:26:55

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment