OEOpen Endeavour open_endeavour.exe
tail -f field_notes.log

Field notes

Dispatches from the lab. Raw, dated, honest — the training runs, the costs, the “why is the loss going UP” moments.

draft

hello, world (or: why train your own embedding model)

The case for doing the unreasonable thing: what embeddings actually are, why owning one matters, and what "one person + one GPU" realistically buys you.

queued

choosing a tokenizer without crying

BPE vs unigram, vocab sizes, and the first of many decisions made in public.

queued

what 768 dimensions feel like

First training run, first benchmark, first invoice. Receipts included.

no newsletter yet. the Discord gets everything first. ¯\_(ツ)_/¯