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
..
aya_300k_lr_summary.csv	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
eval_loss_history.csv	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
eval_loss_over_steps.png	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
grad_norm_over_steps.png	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
gradient_history.csv	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
lr_vs_final_eval_loss.png	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
training_run_summary.csv	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00
training_run_summary.json	Add Aya 300k learning-rate study and benchmark results	2026-10-25 00:33:24 +02:00