Downstream task evaluation after post-training stage SFT for 13b_sd.
Source: source_paper:Table 8, Page 33, After SFT
Reported and observed measurements
downstream_accuracy
t8_sft_13b_sd_gsm8k
Reported 62.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_gsm_s
Reported 42.5 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_gsmplus
Reported 34.8 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_bbh
Reported 31.6 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_drop
Reported 43.5 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_math
Reported 10.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_tqa
Reported 54.9 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_nq
Reported 24.1 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_sqa
Reported 8.5 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_mmlu
Reported 48 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_mmlu_p
Reported 17.7 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_arc_c
Reported 55.1 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_obqa
Reported 59.4 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_wino
Reported 52.2 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_agi
Reported 38 percentage_points
Observed — percentage_points
downstream_accuracy
t8_sft_13b_sd_ife
Reported 43.1 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.