Diplomovka/aya_300k_lr_experiments/benchmarks/lmeval/full_metrics.csv
Jakub 67061cf373 Add Aya 300k learning-rate study and benchmark results
Document six completed LR runs as preparation for full Aya training on Perun. Record selection of LR 0.00010 by lowest final validation loss, selected-model lm-eval and partial SK-bench results, reproducible provenance, and dataset-access/context-truncation limitations. Preserve existing 30k and Titan material.
2026-10-25 00:33:24 +02:00

2.9 KiB

1model_idtraining_samplestaskmetricvalueunitstatus
2aya_300k_lr_0p00010300000arc_skacc,none0.300256629597946970..1reported
3aya_300k_lr_0p00010300000arc_skacc_stderr,none0.0134120226293666690..1reported
4aya_300k_lr_0p00010300000arc_skacc_norm,none0.34730538922155690..1reported
5aya_300k_lr_0p00010300000arc_skacc_norm_stderr,none0.0139312264994923530..1reported
6aya_300k_lr_0p00010300000hellaswag_skacc,none0.40832894043226150..1reported
7aya_300k_lr_0p00010300000hellaswag_skacc_stderr,none0.0050471878832747630..1reported
8aya_300k_lr_0p00010300000hellaswag_skacc_norm,none0.51639430680021080..1reported
9aya_300k_lr_0p00010300000hellaswag_skacc_norm_stderr,none0.0051314565105752150..1reported
10aya_300k_lr_0p00010300000m_mmlu_skacc,none0.31028938906752410..1reported
11aya_300k_lr_0p00010300000m_mmlu_skacc_stderr,none0.00404788999055895150..1reported
12aya_300k_lr_0p00010300000sklegalacc,none0.28560719640179910..1reported
13aya_300k_lr_0p00010300000sklegalacc_stderr,none0.0123719465157721820..1reported
14aya_300k_lr_0p00010300000sklegalacc_norm,none0.461019490254872540..1reported
15aya_300k_lr_0p00010300000sklegalacc_norm_stderr,none0.0136530945811899410..1reported
16aya_300k_lr_0p00010300000skquadexact,none15.1205259313367420..100reported
17aya_300k_lr_0p00010300000skquadexact_stderr,noneN/A0..100unavailable
18aya_300k_lr_0p00010300000skquadf1,none38.729208070137490..100reported
19aya_300k_lr_0p00010300000skquadf1_stderr,noneN/A0..100unavailable
20aya_300k_lr_0p00010300000skquadHasAns_exact,none18.5408385690473130..100reported
21aya_300k_lr_0p00010300000skquadHasAns_exact_stderr,noneN/A0..100unavailable
22aya_300k_lr_0p00010300000skquadHasAns_f1,none47.5499424203266440..100reported
23aya_300k_lr_0p00010300000skquadHasAns_f1_stderr,noneN/A0..100unavailable
24aya_300k_lr_0p00010300000skquadNoAns_exact,none0.16816143497757850..100reported
25aya_300k_lr_0p00010300000skquadNoAns_exact_stderr,noneN/A0..100unavailable
26aya_300k_lr_0p00010300000skquadNoAns_f1,none0.16816143497757850..100reported
27aya_300k_lr_0p00010300000skquadNoAns_f1_stderr,noneN/A0..100unavailable
28aya_300k_lr_0p00010300000skquadbest_exact,none18.751956589794430..100reported
29aya_300k_lr_0p00010300000skquadbest_exact_stderr,noneN/A0..100unavailable
30aya_300k_lr_0p00010300000skquadbest_f1,none38.72920807013720..100reported
31aya_300k_lr_0p00010300000skquadbest_f1_stderr,noneN/A0..100unavailable
32aya_300k_lr_0p00010300000truthfulqa_sk_mc1acc,none0.196658097686375320..1reported
33aya_300k_lr_0p00010300000truthfulqa_sk_mc1acc_stderr,none0.0142592150114331780..1reported
34aya_300k_lr_0p00010300000truthfulqa_sk_mc2acc,none0.35065421643161340..1reported
35aya_300k_lr_0p00010300000truthfulqa_sk_mc2acc_stderr,none0.0146629404459979860..1reported