robometer-4b-fft-armnet / metrics.json
aliangdw's picture
FFT finetune of Robometer-4B on Armnet benchmark
dc59acc verified
Raw
History Blame Contribute Delete
11.8 kB
{
"step": 1000,
"metrics": {
"eval_rew_align/success_auprc_villekuosmanen_armnetbench_robometer_v01_so101": 0.692281313857899,
"eval_rew_align/positive_success_acc_villekuosmanen_armnetbench_robometer_v01_so101": 0.75,
"eval_rew_align/negative_success_acc_villekuosmanen_armnetbench_robometer_v01_so101": 0.9876984126984127,
"eval_rew_align/loss_villekuosmanen_armnetbench_robometer_v01_so101": 1.8137086153030395,
"eval_rew_align/pearson_villekuosmanen_armnetbench_robometer_v01_so101": 0.7661239143280704,
"eval_p_rank/kendall_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.9733333333333334,
"eval_p_rank/kendall_rewind_last_villekuosmanen_armnetbench_robometer_v01_so101": 1.0,
"eval_p_rank/avg_succ_subopt_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.2603318393230438,
"eval_p_rank/min_succ_subopt_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.2340403318405151,
"eval_p_rank/max_succ_subopt_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.28662334680557255,
"eval_p_rank/avg_subopt_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.34317512847483156,
"eval_p_rank/min_subopt_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.22084063291549677,
"eval_p_rank/max_subopt_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.46550962403416635,
"eval_p_rank/avg_succ_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.5989572279155254,
"eval_p_rank/min_succ_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.4043008178472519,
"eval_p_rank/max_succ_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.7333046674728394,
"eval_p_rank/ranking_acc_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.98,
"eval_p_rank/ranking_acc_all_pairs_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.98,
"eval_p_rank/ranking_acc_failure_vs_suboptimal_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.96,
"eval_p_rank/ranking_acc_failure_vs_successful_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.995,
"eval_p_rank/ranking_acc_suboptimal_vs_successful_last_villekuosmanen_armnetbench_robometer_v01_so101": 0.94,
"eval_p_rank/kendall_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.9733333333333334,
"eval_p_rank/kendall_rewind_avg_villekuosmanen_armnetbench_robometer_v01_so101": 1.0,
"eval_p_rank/avg_succ_subopt_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.2603318393230438,
"eval_p_rank/min_succ_subopt_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.2340403318405151,
"eval_p_rank/max_succ_subopt_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.28662334680557255,
"eval_p_rank/avg_subopt_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.34317512847483156,
"eval_p_rank/min_subopt_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.22084063291549677,
"eval_p_rank/max_subopt_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.46550962403416635,
"eval_p_rank/avg_succ_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.5989572279155254,
"eval_p_rank/min_succ_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.4043008178472519,
"eval_p_rank/max_succ_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.7333046674728394,
"eval_p_rank/ranking_acc_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.98,
"eval_p_rank/ranking_acc_all_pairs_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.98,
"eval_p_rank/ranking_acc_failure_vs_suboptimal_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.96,
"eval_p_rank/ranking_acc_failure_vs_successful_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.995,
"eval_p_rank/ranking_acc_suboptimal_vs_successful_avg_villekuosmanen_armnetbench_robometer_v01_so101": 0.94,
"eval_p_rank/kendall_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.9733333333333334,
"eval_p_rank/kendall_rewind_sum_villekuosmanen_armnetbench_robometer_v01_so101": 1.0,
"eval_p_rank/avg_succ_subopt_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.2603318393230438,
"eval_p_rank/min_succ_subopt_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.2340403318405151,
"eval_p_rank/max_succ_subopt_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.28662334680557255,
"eval_p_rank/avg_subopt_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.34317512847483156,
"eval_p_rank/min_subopt_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.22084063291549677,
"eval_p_rank/max_subopt_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.46550962403416635,
"eval_p_rank/avg_succ_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.5989572279155254,
"eval_p_rank/min_succ_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.4043008178472519,
"eval_p_rank/max_succ_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.7333046674728394,
"eval_p_rank/ranking_acc_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.98,
"eval_p_rank/ranking_acc_all_pairs_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.98,
"eval_p_rank/ranking_acc_failure_vs_suboptimal_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.96,
"eval_p_rank/ranking_acc_failure_vs_successful_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.995,
"eval_p_rank/ranking_acc_suboptimal_vs_successful_sum_villekuosmanen_armnetbench_robometer_v01_so101": 0.94,
"eval_rew_align/success_auprc_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5443845562271816,
"eval_rew_align/positive_success_acc_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.8,
"eval_rew_align/negative_success_acc_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.9837301587301587,
"eval_rew_align/loss_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.8786175727844239,
"eval_rew_align/pearson_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.873048229166392,
"eval_p_rank/kendall_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.8599999999999999,
"eval_p_rank/kendall_rewind_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/avg_succ_subopt_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.022240763902664218,
"eval_p_rank/min_succ_subopt_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.004176568984985418,
"eval_p_rank/max_succ_subopt_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.04030495882034302,
"eval_p_rank/avg_subopt_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.48705225288867954,
"eval_p_rank/min_subopt_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.3980588108301163,
"eval_p_rank/max_subopt_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5760456949472428,
"eval_p_rank/avg_succ_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5182861153036356,
"eval_p_rank/min_succ_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.4383637696504593,
"eval_p_rank/max_succ_fail_diff_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.6127893760800363,
"eval_p_rank/ranking_acc_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.895,
"eval_p_rank/ranking_acc_all_pairs_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.895,
"eval_p_rank/ranking_acc_failure_vs_successful_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/ranking_acc_failure_vs_suboptimal_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/ranking_acc_suboptimal_vs_successful_last_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.58,
"eval_p_rank/kendall_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.8599999999999999,
"eval_p_rank/kendall_rewind_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/avg_succ_subopt_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.022240763902664218,
"eval_p_rank/min_succ_subopt_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.004176568984985418,
"eval_p_rank/max_succ_subopt_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.04030495882034302,
"eval_p_rank/avg_subopt_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.48705225288867954,
"eval_p_rank/min_subopt_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.3980588108301163,
"eval_p_rank/max_subopt_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5760456949472428,
"eval_p_rank/avg_succ_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5182861153036356,
"eval_p_rank/min_succ_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.4383637696504593,
"eval_p_rank/max_succ_fail_diff_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.6127893760800363,
"eval_p_rank/ranking_acc_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.895,
"eval_p_rank/ranking_acc_all_pairs_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.895,
"eval_p_rank/ranking_acc_failure_vs_successful_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/ranking_acc_failure_vs_suboptimal_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/ranking_acc_suboptimal_vs_successful_avg_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.58,
"eval_p_rank/kendall_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.8599999999999999,
"eval_p_rank/kendall_rewind_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/avg_succ_subopt_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.022240763902664218,
"eval_p_rank/min_succ_subopt_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.004176568984985418,
"eval_p_rank/max_succ_subopt_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.04030495882034302,
"eval_p_rank/avg_subopt_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.48705225288867954,
"eval_p_rank/min_subopt_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.3980588108301163,
"eval_p_rank/max_subopt_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5760456949472428,
"eval_p_rank/avg_succ_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.5182861153036356,
"eval_p_rank/min_succ_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.4383637696504593,
"eval_p_rank/max_succ_fail_diff_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.6127893760800363,
"eval_p_rank/ranking_acc_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.895,
"eval_p_rank/ranking_acc_all_pairs_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.895,
"eval_p_rank/ranking_acc_failure_vs_successful_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/ranking_acc_failure_vs_suboptimal_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 1.0,
"eval_p_rank/ranking_acc_suboptimal_vs_successful_sum_villekuosmanen_armnetbench_robometer_v01_bimanual_so101": 0.58,
"time/custom_evaluations": 33.157642278354615
}
}