Downstream task evaluation after post-training stage SFT for 7b_fkd.
Source: source_paper:Table 8, Page 33, After SFT
Reported and observed measurements
downstream_accuracy
t8_sft_7b_fkd_gsm8k
Reported 54.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_gsm_s
Reported 34.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_gsmplus
Reported 30.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_bbh
Reported 31.3 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_drop
Reported 39.1 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_math
Reported 9.4 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_tqa
Reported 54.1 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_nq
Reported 23.3 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_sqa
Reported 8.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_mmlu
Reported 49.1 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_mmlu_p
Reported 18.5 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_arc_c
Reported 61.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_obqa
Reported 59.4 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_wino
Reported 51.9 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_agi
Reported 39.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_7b_fkd_ife
Reported 46 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.