Skip to content

Instantly share code, notes, and snippets.

@mlabonne
Created February 25, 2024 01:37
Show Gist options
  • Save mlabonne/c8c0914945f9c189cca74120bc834c3e to your computer and use it in GitHub Desktop.
Save mlabonne/c8c0914945f9c189cca74120bc834c3e to your computer and use it in GitHub Desktop.
Model AGIEval GPT4All TruthfulQA Bigbench Average
OrcaGemma-2B 24.44 42.49 45.84 29.76 35.63

AGIEval

Task Version Metric Value Stderr
agieval_aqua_rat 0 acc 23.62 ± 2.67
acc_norm 22.05 ± 2.61
agieval_logiqa_en 0 acc 22.43 ± 1.64
acc_norm 27.04 ± 1.74
agieval_lsat_ar 0 acc 22.17 ± 2.75
acc_norm 19.13 ± 2.60
agieval_lsat_lr 0 acc 16.86 ± 1.66
acc_norm 21.57 ± 1.82
agieval_lsat_rc 0 acc 24.54 ± 2.63
acc_norm 24.16 ± 2.61
agieval_sat_en 0 acc 31.07 ± 3.23
acc_norm 32.52 ± 3.27
agieval_sat_en_without_passage 0 acc 27.18 ± 3.11
acc_norm 28.16 ± 3.14
agieval_sat_math 0 acc 23.64 ± 2.87
acc_norm 20.91 ± 2.75

Average: 24.44%

GPT4All

Task Version Metric Value Stderr
arc_challenge 0 acc 21.76 ± 1.21
acc_norm 25.17 ± 1.27
arc_easy 0 acc 38.97 ± 1.00
acc_norm 37.08 ± 0.99
boolq 1 acc 65.47 ± 0.83
hellaswag 0 acc 31.26 ± 0.46
acc_norm 36.50 ± 0.48
openbookqa 0 acc 17.20 ± 1.69
acc_norm 27.60 ± 2.00
piqa 0 acc 57.24 ± 1.15
acc_norm 56.69 ± 1.16
winogrande 0 acc 48.93 ± 1.40

Average: 42.49%

TruthfulQA

Task Version Metric Value Stderr
truthfulqa_mc 1 mc1 27.17 ± 1.56
mc2 45.84 ± 1.55

Average: 45.84%

Bigbench

Task Version Metric Value Stderr
bigbench_causal_judgement 0 multiple_choice_grade 50.53 ± 3.64
bigbench_date_understanding 0 multiple_choice_grade 25.20 ± 2.26
bigbench_disambiguation_qa 0 multiple_choice_grade 31.40 ± 2.89
bigbench_geometric_shapes 0 multiple_choice_grade 3.34 ± 0.95
exact_str_match 0.00 ± 0.00
bigbench_logical_deduction_five_objects 0 multiple_choice_grade 25.80 ± 1.96
bigbench_logical_deduction_seven_objects 0 multiple_choice_grade 21.43 ± 1.55
bigbench_logical_deduction_three_objects 0 multiple_choice_grade 41.00 ± 2.84
bigbench_movie_recommendation 0 multiple_choice_grade 24.80 ± 1.93
bigbench_navigate 0 multiple_choice_grade 51.10 ± 1.58
bigbench_reasoning_about_colored_objects 0 multiple_choice_grade 26.15 ± 0.98
bigbench_ruin_names 0 multiple_choice_grade 21.43 ± 1.94
bigbench_salient_translation_error_detection 0 multiple_choice_grade 12.93 ± 1.06
bigbench_snarks 0 multiple_choice_grade 51.38 ± 3.73
bigbench_sports_understanding 0 multiple_choice_grade 49.70 ± 1.59
bigbench_temporal_sequences 0 multiple_choice_grade 27.20 ± 1.41
bigbench_tracking_shuffled_objects_five_objects 0 multiple_choice_grade 18.08 ± 1.09
bigbench_tracking_shuffled_objects_seven_objects 0 multiple_choice_grade 13.26 ± 0.81
bigbench_tracking_shuffled_objects_three_objects 0 multiple_choice_grade 41.00 ± 2.84

Average: 29.76%

Average score: 35.63%

Elapsed time: 02:39:22

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