H1K0 d9585ec008 data: add fine-tuning run results (lr=3e-5, 30 epochs) ...

val loss 1.19 → 0.77, val perplexity 3.29 → 2.15.
Best epoch 20, early stop at epoch 30 (patience=10).
Improvement over previous lr=1e-5 run (best val ppl 2.22).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-05-21 20:52:39 +03:00

..

_smoke_pretrain.log.csv

data: add 2-epoch smoke pre-training log (_smoke_pretrain.log.csv)

2026-05-20 12:16:28 +03:00

.gitkeep

chore: initialize project scaffold

2026-05-19 10:28:17 +03:00

finetuned_curves.png

data: add fine-tuning run results (lr=3e-5, 30 epochs)

2026-05-21 20:52:39 +03:00

finetuned.log.csv

data: add fine-tuning run results (lr=3e-5, 30 epochs)

2026-05-21 20:52:39 +03:00

finetuned.report.txt

data: add fine-tuning run results (lr=3e-5, 30 epochs)

2026-05-21 20:52:39 +03:00

pretrained_curves.png

data: update pretrained checkpoint results (BAR-free tokenizer)

2026-05-20 14:28:00 +03:00

pretrained.log.csv

data: update pretrained checkpoint results (BAR-free tokenizer)

2026-05-20 14:28:00 +03:00

pretrained.report.txt

data: update pretrained checkpoint results (BAR-free tokenizer)

2026-05-20 14:28:00 +03:00