Per-task downstream accuracy for mid-training ablation teacher_correct using OLMo-2 7B Instruct teacher.
Source: source_paper:Table 9, Page 33
Reported and observed measurements
downstream_accuracy
t9_teacher_correct_gsm8k
Reported 63.8 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_gsm_s
Reported 49.5 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_gsmplus
Reported 41.7 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_bbh
Reported 31.8 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_drop
Reported 45.2 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_math
Reported 9.8 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_tqa
Reported 44.5 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_nq
Reported 19.4 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_sqa
Reported 8.7 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_mmlu
Reported 50.6 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_mmlu_p
Reported 19.3 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_arc_c
Reported 63.2 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_obqa
Reported 64.6 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_wino
Reported 52 percentage_points
Observed — percentage_points
downstream_accuracy
t9_teacher_correct_agi
Reported 39.9 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.