Previous direct/ had 'I feel X' first-person descriptions. The
training run showed they formed their own format-cluster: all 7
concepts leaned into the same 5-6 dims (d2455, d505, d2955,
d1236) with negative sign, while the 91 story-based concepts
leaned into those dims with positive sign. PCA found the
direct-vs-narrative format axis as a major variance direction,
isolating the 7 concepts in their own island.
Rewrite as 3rd-person narrative stories matching the rest of
the corpus. Keeps the explicit anchor phrases that worked ('it
all clicked into place', 'she was terrified', 'it was
anticipatory grief') but drops the first-person 'I feel X'
that was the format signal.
Each of the 7 concepts now has 3 narrative stories in varied
settings (conversations, drives, kitchens, mothers+grandmothers,
work, investigations). The blank-line-separated format is
still loaded by _load_direct_descriptions.
Also drop _baseline.txt — it was first-person ('I feel fine.
...') and would re-introduce the format mismatch. The ~90
story-based concepts provide plenty of narrative negatives
for each concept's training.
5 lines
768 B
Text
5 lines
768 B
Text
She'd been working through the problem, sifting through all the disparate parts — then suddenly, it all made sense, it all clicked into place. The pieces arranged themselves in a single motion. She sat there, catching up to what she'd already seen.
|
|
|
|
He'd been listening to her for twenty minutes, something not adding up in the way she was telling it — then suddenly, it all clicked into place. The gap wasn't an accident. She was telling him two different stories and hoping he wouldn't notice. He saw the whole shape of it.
|
|
|
|
She'd been reading the old letter without understanding what her father meant — then suddenly, it all made sense. He hadn't been warning her. He'd been telling her he was leaving. Twenty years later, in her kitchen, it finally landed.
|