Updated 01 Oct 2026
Now
What I’m working on, and the next questions to answer.
01Research
Alethic Phase I
~1B target, 8,240-token context, and a controlled matched baseline.
SGCA
Testing the separation between attention retrieval and residual control.
02Building
Dataset and training infrastructure
Export, source-ID deduplication, packed binaries, and resumable preparation.
03Learning
Efficient language-model architecture, experiment design, and the systems underneath training.
04Publishing next
A complete run record: actual configurations, tokenizer and data documentation, checkpoints, logs, and measured comparisons when available.
A manually maintained snapshot, last updated 01 Oct 2026.