Skip to content

Instantly share code, notes, and snippets.

@mlabonne
Created March 21, 2024 22:58
Show Gist options
  • Save mlabonne/e7ef034968c58c666493d9236473ceee to your computer and use it in GitHub Desktop.
Save mlabonne/e7ef034968c58c666493d9236473ceee to your computer and use it in GitHub Desktop.
Model AGIEval GPT4All TruthfulQA Bigbench Average
UltraMerge-7B 44.36 77.15 78.47 49.35 62.33

AGIEval

Task Version Metric Value Stderr
agieval_aqua_rat 0 acc 27.56 ± 2.81
acc_norm 23.23 ± 2.65
agieval_logiqa_en 0 acc 39.48 ± 1.92
acc_norm 39.63 ± 1.92
agieval_lsat_ar 0 acc 23.91 ± 2.82
acc_norm 23.48 ± 2.80
agieval_lsat_lr 0 acc 53.73 ± 2.21
acc_norm 54.12 ± 2.21
agieval_lsat_rc 0 acc 65.06 ± 2.91
acc_norm 65.43 ± 2.91
agieval_sat_en 0 acc 77.18 ± 2.93
acc_norm 77.18 ± 2.93
agieval_sat_en_without_passage 0 acc 42.72 ± 3.45
acc_norm 42.23 ± 3.45
agieval_sat_math 0 acc 31.82 ± 3.15
acc_norm 29.55 ± 3.08

Average: 44.36%

GPT4All

Task Version Metric Value Stderr
arc_challenge 0 acc 67.06 ± 1.37
acc_norm 67.66 ± 1.37
arc_easy 0 acc 86.32 ± 0.71
acc_norm 79.92 ± 0.82
boolq 1 acc 87.25 ± 0.58
hellaswag 0 acc 70.08 ± 0.46
acc_norm 87.57 ± 0.33
openbookqa 0 acc 40.40 ± 2.20
acc_norm 50.20 ± 2.24
piqa 0 acc 83.30 ± 0.87
acc_norm 85.36 ± 0.82
winogrande 0 acc 82.08 ± 1.08

Average: 77.15%

TruthfulQA

Task Version Metric Value Stderr
truthfulqa_mc 1 mc1 62.91 ± 1.69
mc2 78.47 ± 1.36

Average: 78.47%

Bigbench

Task Version Metric Value Stderr
bigbench_causal_judgement 0 multiple_choice_grade 60.00 ± 3.56
bigbench_date_understanding 0 multiple_choice_grade 61.52 ± 2.54
bigbench_disambiguation_qa 0 multiple_choice_grade 52.71 ± 3.11
bigbench_geometric_shapes 0 multiple_choice_grade 22.84 ± 2.22
exact_str_match 0.00 ± 0.00
bigbench_logical_deduction_five_objects 0 multiple_choice_grade 32.40 ± 2.10
bigbench_logical_deduction_seven_objects 0 multiple_choice_grade 24.29 ± 1.62
bigbench_logical_deduction_three_objects 0 multiple_choice_grade 58.33 ± 2.85
bigbench_movie_recommendation 0 multiple_choice_grade 54.00 ± 2.23
bigbench_navigate 0 multiple_choice_grade 57.20 ± 1.57
bigbench_reasoning_about_colored_objects 0 multiple_choice_grade 68.70 ± 1.04
bigbench_ruin_names 0 multiple_choice_grade 56.47 ± 2.35
bigbench_salient_translation_error_detection 0 multiple_choice_grade 38.38 ± 1.54
bigbench_snarks 0 multiple_choice_grade 74.03 ± 3.27
bigbench_sports_understanding 0 multiple_choice_grade 75.25 ± 1.37
bigbench_temporal_sequences 0 multiple_choice_grade 52.60 ± 1.58
bigbench_tracking_shuffled_objects_five_objects 0 multiple_choice_grade 22.40 ± 1.18
bigbench_tracking_shuffled_objects_seven_objects 0 multiple_choice_grade 18.91 ± 0.94
bigbench_tracking_shuffled_objects_three_objects 0 multiple_choice_grade 58.33 ± 2.85

Average: 49.35%

Average score: 62.33%

Elapsed time: 02:15:54

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