Full downstream results after standard 4-stage post-training pipeline (SFT, DPO, RLVR1, RLVR2) for 7B Switch Distillation post-training performance.
Source: source_paper:Table 2, Page 9
Reported and observed measurements
downstream_accuracy
t2_7b_sd_gsm8k
Reported 79.8 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_gsm_s
Reported 65 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_gsmplus
Reported 52.7 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_bbh
Reported 35.6 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_drop
Reported 48.2 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_math
Reported 22.4 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_tqa
Reported 53.6 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_nq
Reported 22.7 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_sqa
Reported 8.2 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_mmlu
Reported 48.1 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_mmlu_p
Reported 19.8 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_arc_c
Reported 62.6 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_obqa
Reported 59.4 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_wino
Reported 54.6 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_agi
Reported 39.6 percentage_points
Observed — percentage_points
downstream_accuracy
t2_7b_sd_ife
Reported 69.5 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.