Skip to content

Instantly share code, notes, and snippets.

@mlabonne
Forked from CultriX-Github/NeuralTrix-bf16-Nous.md
Created February 14, 2024 19:53
Show Gist options
  • Save mlabonne/1e7d6f17fbd2fd5ce34079b25edcc06f to your computer and use it in GitHub Desktop.
Save mlabonne/1e7d6f17fbd2fd5ce34079b25edcc06f to your computer and use it in GitHub Desktop.
Model AGIEval GPT4All TruthfulQA Bigbench Average
NeuralTrix-bf16 44.43 76.43 80.18 49.23 62.57

AGIEval

Task Version Metric Value Stderr
agieval_aqua_rat 0 acc 25.59 ± 2.74
acc_norm 25.20 ± 2.73
agieval_logiqa_en 0 acc 37.48 ± 1.90
acc_norm 37.33 ± 1.90
agieval_lsat_ar 0 acc 23.91 ± 2.82
acc_norm 24.35 ± 2.84
agieval_lsat_lr 0 acc 53.33 ± 2.21
acc_norm 54.31 ± 2.21
agieval_lsat_rc 0 acc 66.54 ± 2.88
acc_norm 65.80 ± 2.90
agieval_sat_en 0 acc 77.67 ± 2.91
acc_norm 78.16 ± 2.89
agieval_sat_en_without_passage 0 acc 42.23 ± 3.45
acc_norm 40.78 ± 3.43
agieval_sat_math 0 acc 32.27 ± 3.16
acc_norm 29.55 ± 3.08

Average: 44.43%

GPT4All

Task Version Metric Value Stderr
arc_challenge 0 acc 65.87 ± 1.39
acc_norm 66.89 ± 1.38
arc_easy 0 acc 84.85 ± 0.74
acc_norm 77.65 ± 0.85
boolq 1 acc 87.06 ± 0.59
hellaswag 0 acc 69.69 ± 0.46
acc_norm 87.44 ± 0.33
openbookqa 0 acc 40.60 ± 2.20
acc_norm 50.00 ± 2.24
piqa 0 acc 82.59 ± 0.88
acc_norm 83.84 ± 0.86
winogrande 0 acc 82.08 ± 1.08

Average: 76.43%

TruthfulQA

Task Version Metric Value Stderr
truthfulqa_mc 1 mc1 63.89 ± 1.68
mc2 80.18 ± 1.33

Average: 80.18%

Bigbench

Task Version Metric Value Stderr
bigbench_causal_judgement 0 multiple_choice_grade 60.53 ± 3.56
bigbench_date_understanding 0 multiple_choice_grade 59.62 ± 2.56
bigbench_disambiguation_qa 0 multiple_choice_grade 59.69 ± 3.06
bigbench_geometric_shapes 0 multiple_choice_grade 22.56 ± 2.21
exact_str_match 0.00 ± 0.00
bigbench_logical_deduction_five_objects 0 multiple_choice_grade 31.80 ± 2.08
bigbench_logical_deduction_seven_objects 0 multiple_choice_grade 22.14 ± 1.57
bigbench_logical_deduction_three_objects 0 multiple_choice_grade 59.67 ± 2.84
bigbench_movie_recommendation 0 multiple_choice_grade 51.60 ± 2.24
bigbench_navigate 0 multiple_choice_grade 53.30 ± 1.58
bigbench_reasoning_about_colored_objects 0 multiple_choice_grade 66.65 ± 1.05
bigbench_ruin_names 0 multiple_choice_grade 56.03 ± 2.35
bigbench_salient_translation_error_detection 0 multiple_choice_grade 38.98 ± 1.54
bigbench_snarks 0 multiple_choice_grade 72.93 ± 3.31
bigbench_sports_understanding 0 multiple_choice_grade 74.44 ± 1.39
bigbench_temporal_sequences 0 multiple_choice_grade 53.20 ± 1.58
bigbench_tracking_shuffled_objects_five_objects 0 multiple_choice_grade 23.60 ± 1.20
bigbench_tracking_shuffled_objects_seven_objects 0 multiple_choice_grade 19.71 ± 0.95
bigbench_tracking_shuffled_objects_three_objects 0 multiple_choice_grade 59.67 ± 2.84

Average: 49.23%

Average score: 62.57%

Elapsed time: 01:42:42

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