DevQuasar/analytical_reasoning_Llama-3-8B

'Make knowledge free for everyone'

Model

Unsloth 4bit qunatized Llama3 8B Fine-Tuned for Analytical reasoning.

Eval

Tasks	Version	Filter	n-shot	Metric		Value		Stderr
hellaswag	1	none	0	acc	↑	0.5704	±	0.0049
		none	0	acc_norm	↑	0.7526	±	0.0043
leaderboard_bbh	N/A
- leaderboard_bbh_boolean_expressions	1	none	3	acc_norm	↑	0.7480	±	0.0275
- leaderboard_bbh_causal_judgement	1	none	3	acc_norm	↑	0.6524	±	0.0349
- leaderboard_bbh_date_understanding	1	none	3	acc_norm	↑	0.4400	±	0.0315
- leaderboard_bbh_disambiguation_qa	1	none	3	acc_norm	↑	0.6240	±	0.0307
- leaderboard_bbh_formal_fallacies	1	none	3	acc_norm	↑	0.5440	±	0.0316
- leaderboard_bbh_geometric_shapes	1	none	3	acc_norm	↑	0.3160	±	0.0295
- leaderboard_bbh_hyperbaton	1	none	3	acc_norm	↑	0.4920	±	0.0317
- leaderboard_bbh_logical_deduction_five_objects	1	none	3	acc_norm	↑	0.4040	±	0.0311
- leaderboard_bbh_logical_deduction_seven_objects	1	none	3	acc_norm	↑	0.4200	±	0.0313
- leaderboard_bbh_logical_deduction_three_objects	1	none	3	acc_norm	↑	0.5360	±	0.0316
- leaderboard_bbh_movie_recommendation	1	none	3	acc_norm	↑	0.5920	±	0.0311
- leaderboard_bbh_navigate	1	none	3	acc_norm	↑	0.5000	±	0.0317
- leaderboard_bbh_object_counting	1	none	3	acc_norm	↑	0.4720	±	0.0316
- leaderboard_bbh_penguins_in_a_table	1	none	3	acc_norm	↑	0.5274	±	0.0415
- leaderboard_bbh_reasoning_about_colored_objects	1	none	3	acc_norm	↑	0.5480	±	0.0315
- leaderboard_bbh_ruin_names	1	none	3	acc_norm	↑	0.5640	±	0.0314
- leaderboard_bbh_salient_translation_error_detection	1	none	3	acc_norm	↑	0.4640	±	0.0316
- leaderboard_bbh_snarks	1	none	3	acc_norm	↑	0.4831	±	0.0376
- leaderboard_bbh_sports_understanding	1	none	3	acc_norm	↑	0.6200	±	0.0308
- leaderboard_bbh_temporal_sequences	1	none	3	acc_norm	↑	0.1720	±	0.0239
- leaderboard_bbh_tracking_shuffled_objects_five_objects	1	none	3	acc_norm	↑	0.2000	±	0.0253
- leaderboard_bbh_tracking_shuffled_objects_seven_objects	1	none	3	acc_norm	↑	0.1840	±	0.0246
- leaderboard_bbh_tracking_shuffled_objects_three_objects	1	none	3	acc_norm	↑	0.3440	±	0.0301
- leaderboard_bbh_web_of_lies	1	none	3	acc_norm	↑	0.5280	±	0.0316
leaderboard_gpqa	N/A
- leaderboard_gpqa_diamond	1	none	0	acc_norm	↑	0.2727	±	0.0317
- leaderboard_gpqa_extended	1	none	0	acc_norm	↑	0.3095	±	0.0198
- leaderboard_gpqa_main	1	none	0	acc_norm	↑	0.3058	±	0.0218
leaderboard_ifeval	3	none	0	inst_level_loose_acc	↑	0.6031	±	N/A
		none	0	inst_level_strict_acc	↑	0.5132	±	N/A
		none	0	prompt_level_loose_acc	↑	0.4677	±	0.0215
		none	0	prompt_level_strict_acc	↑	0.3660	±	0.0207
leaderboard_math_hard	N/A
- leaderboard_math_algebra_hard	2	none	4	exact_match	↑	0.1303	±	0.0192
- leaderboard_math_counting_and_prob_hard	2	none	4	exact_match	↑	0.0244	±	0.0140
- leaderboard_math_geometry_hard	2	none	4	exact_match	↑	0.0303	±	0.0150
- leaderboard_math_intermediate_algebra_hard	2	none	4	exact_match	↑	0.0179	±	0.0079
- leaderboard_math_num_theory_hard	2	none	4	exact_match	↑	0.0714	±	0.0208
- leaderboard_math_prealgebra_hard	2	none	4	exact_match	↑	0.1451	±	0.0254
- leaderboard_math_precalculus_hard	2	none	4	exact_match	↑	0.0296	±	0.0146
leaderboard_mmlu_pro	0.1	none	5	acc	↑	0.3481	±	0.0043
leaderboard_musr	N/A
- leaderboard_musr_murder_mysteries	1	none	0	acc_norm	↑	0.5480	±	0.0315
- leaderboard_musr_object_placements	1	none	0	acc_norm	↑	0.2617	±	0.0275
- leaderboard_musr_team_allocation	1	none	0	acc_norm	↑	0.3640	±	0.0305

Made with