Per-task downstream accuracy for mid-training ablation sd_fkl using OLMo-2 7B Instruct teacher.
Source: source_paper:Table 9, Page 33
Reported and observed measurements
downstream_accuracy
t9_sd_fkl_gsm8k
Reported 66.7 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_gsm_s
Reported 52 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_gsmplus
Reported 43.7 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_bbh
Reported 31.2 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_drop
Reported 46.4 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_math
Reported 11 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_tqa
Reported 54.8 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_nq
Reported 24.3 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_sqa
Reported 8.2 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_mmlu
Reported 50.5 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_mmlu_p
Reported 19.1 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_arc_c
Reported 63.7 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_obqa
Reported 62.6 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_wino
Reported 52.2 percentage_points
Observed — percentage_points
downstream_accuracy
t9_sd_fkl_agi
Reported 39.6 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.