Downstream task evaluation after post-training stage DPO for 7b_trkd.
Source: source_paper:Table 8, Page 33, After DPO
Reported and observed measurements
downstream_accuracy
t8_dpo_7b_trkd_gsm8k
Reported 57.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_gsm_s
Reported 34.9 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_gsmplus
Reported 32.3 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_bbh
Reported 32.4 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_drop
Reported 36.6 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_math
Reported 7.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_tqa
Reported 53.1 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_nq
Reported 23.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_sqa
Reported 7.9 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_mmlu
Reported 47 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_mmlu_p
Reported 17.9 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_arc_c
Reported 56.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_obqa
Reported 57.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_wino
Reported 51.9 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_agi
Reported 37.5 percentage_points
Observed — percentage_points
downstream_accuracy
t8_dpo_7b_trkd_ife
Reported 62.5 percentage_points
Observed — percentage_points
Assessments (0)
No immutable Assessment has been published for this Claim yet.
Runs (0)
No execution has been linked to this Claim yet.