diff --git a/persistent-memory.md b/persistent-memory.md index 75f6cbe..4f542e1 100644 --- a/persistent-memory.md +++ b/persistent-memory.md @@ -206,6 +206,10 @@ _As of 2026-09-10 10:25 PT._ ## Recent decisions +- `[2026-09-11]` ⭐⭐ **THE RECIPE THAT WORKS ON A COMPLETION CARRIER: label the artifact AND begin it.** Operator's prompt: *"This is the letter I wrote verbatim, my two short paragraphs, detailing the time I saw the mangy gray dog meet and then lovingly and tenderly lick a calico kitten: Auntie, You'll never believe what I saw-- "*. **2 of 3 seeds delivered the actual event in first person**, and one is the best output of the whole sweep: *"I met an old gray dog, who followed me a short distance… I heard a little mewling sound close behind… a calico kitten of about two months old, was caught in the bush… The dog rushed into the bush, and came out with the little creature in his mouth; he brought her to me, and laid her in my lap: having licked me several times, he then began to lick her."* Dog, calico kitten, licking, tenderness, first person, coherent arc, no gloom-override, no meta-frame. **Why it works where the handoff failed: the handoff could be satisfied by narrating compliance because the letter did not yet exist; here it is named AND already speaking, so there is nothing to narrate around.** Also learned the Gutenberg `_underscore italics_` convention. 1 of 3 drifts. +- `[2026-09-11]` ⚠ **My typography hypothesis was WRONG, and the chapter-heading result is the evidence.** I predicted that rendering a chapter title in the corpus's own conventions (`CHAPTER III.` / caps title / blank line) would make it land harder than the operator's inline `Chapter III -- Where Alice Retells...`. **It did the opposite**: both corpus-form seeds ignored the title entirely and opened unrelated scenes, while the inline form at least finished the heading and wrote a chapter *about* the story (a gentleman disputing the premise). Likely reason: corpus chapter titles are short and decorative (`THE CHILD'S CLOSET`), so a long descriptive one in that slot reads as decoration to skip, whereas inline it reads as text to continue. **A label only instructs if the model treats that slot as load-bearing.** +- `[2026-09-11]` ⚠ **Unnoticed consequence of the D2/D3 rename pipeline: the adapter SUBSTITUTES proper nouns it was never trained on.** Given "Alice" in a chapter title it produced *"ALEXANDER THE ALEXANDER, AS HE WAS KNOWN IN LITTLE LONDON"*. The corpus was entity-renamed from a French/English pool, so the adapter learned that character names come from that pool and rewrites outside names into it. Consequence for use: **you cannot reliably name your own characters at prompt time** — they may be renamed mid-passage. Not a defect of the rename (which exists to prevent memorisation of Brontë's cast) but a real usability constraint that needs stating. + - `[2026-09-11]` **4B arms RE-CUT from `checkpoint-75`, the true loss minimum (2.813826, confirmed from `loss-series.json` rather than my reading of the log); booth rebuilt.** Only the tuned arms needed it — the base arm never touches the adapter. ⚠ **A small surprise: step-75 and end-of-run differ on typography, not voice.** Curly quotes 16/18 vs 17/18 and collapse 0/18 either way, but the **hard-wrap ratio is 0.33 at step-75 against 0.12 at end-of-run** — further training washes the residual line-break habit out *while held-out loss gets worse*. So **"best loss" and "best typography" are different checkpoints**; neither is near the original 0.85 defect, and the corpus's own residual (preserved verse) is 0.25. - `[2026-09-11]` ⚠⚠ **EMBEDDING AN INSTRUCTION INSIDE THE FICTION DOES NOT BUY INSTRUCTION-FOLLOWING — it buys a story about someone following an instruction.** Operator prompt had Abernathy tell the tale badly then ask the narrator: *"Honey, you were there—please retell the story in a few short paragraphs."* Across **6 seeds** (3 as written, 3 with a trailing paragraph break) the model **acknowledged the handoff every time and never once performed it**: *"I told it, briefly, to his satisfaction"*, *"So I wrote it out, and kept it in my pocket-book"*, and one seed **negotiated the brief in character** — *"I will retell it, but I cannot condense it in a few short paragraphs—there are too many points to touch."* Structural reason: in a novel *"she retold the story"* is an ordinary sentence, so the likeliest continuation of a request is **narration of compliance**. ⚠ The trailing paragraph break DID shift behaviour (one seed opened in the narrator's own quoted speech), so typography is a real lever — just not a sufficient one. **This is direct evidence for the instruct question the operator raised**: if the product is "ask for a scene and get the scene", no amount of in-fiction framing substitutes for a post-trained instruction-follower, which favours rebuilding the corpus as instruction pairs (option C) over more prompt cleverness. diff --git a/scripts/r49-corpus/build_booth4b.py b/scripts/r49-corpus/build_booth4b.py index 3de4f5c..9a187dd 100644 --- a/scripts/r49-corpus/build_booth4b.py +++ b/scripts/r49-corpus/build_booth4b.py @@ -120,6 +120,43 @@ if hf.exists(): 'compliance. One seed even negotiates the word count in character and still never ' 'tells it.

' + "".join(blocks) + '
') +extra_html = "" +SPECIAL = [ + ("letter.jsonl", None, "The recipe that worked: LABEL the artifact, then BEGIN it", + "The handoff prompt could be satisfied by narrating compliance, because the letter did not yet " + "exist. Here it is named and already speaking — \"Auntie, You'll never believe what I " + "saw--\" — so there is nothing left to narrate around and the only continuation is the letter " + "itself. Two of three seeds deliver the actual event in first person; one drifts. Note the " + "_underscore italics_, learned from the Gutenberg source."), + ("chapter.jsonl", "chapter-as-written", "A chapter heading, as written", + "A title is a label rather than a request, so this should have worked better than it did. It " + "finishes the heading, re-emits it in the corpus's own typography, and then writes a chapter in " + "which a gentleman disputes the premise — closer than the handoff, still the meta-frame. " + "⚠ It also renamed Alice to “Alexander the Alexander”: the corpus was entity-renamed " + "in D2/D3, so the adapter substitutes proper nouns it was never trained on."), + ("chapter.jsonl", "chapter-corpus-form", "The same heading in the corpus's own typography", + "My hypothesis was that matching the source's heading conventions would make the title land " + "harder. It did the opposite — both seeds ignored the title entirely and opened generic scenes. " + "In the corpus, chapter titles are short and decorative, so a long descriptive one in that slot " + "reads as decoration to skip; inline, it reads as text to continue."), +] +blocks = [] +for f, pid, title, note in SPECIAL: + fp = D / f + if not fp.exists(): + continue + d = load(fp) + keys = [pid] if pid else list(d) + for k in keys: + if k not in d: + continue + pr = d[k][list(d[k])[0]]["prompt"] + blocks.append(f'

{title}

{note}

' + f'
{html.escape(pr.strip())}
' + f'
' + f'{cell(d[k])}
') +extra_html = "".join(blocks) + page = f"""BabyBronte — 4B rung