Replace clustering with a catalogue, and hand-write the references to judge it against
The pipeline's embed-and-cluster step is dead, and this commit holds both the evidence for that and the step proposed to replace it. Predicaments. Scenes are re-described as "what the person is up against", with no names, jobs or places, then embedded and clustered (redescribe.py, topic_words.py, topic_share.py). The pilot chose qwen3:14b over 3b by reading both side by side. Two defects the pilot exposed are fixed: split.py missed titles in quotes and a contents subtitle after a dash, so three stories had been merged into their neighbours, and strip_names.py read New York place names as people. The corrected corpus is probe/v2 (97 stories, 839 scenes); carry_summaries.py reuses the 829 unchanged v1 summaries. Topic share fell from 20% to 13% at k=60, short of the pre-registered 10%. Hand references. Three corpora were read scene by scene and written up by hand, under the same prompt rules the local models get, as a baseline to judge them against: O. Henry (probe/v2/claude, 839 scenes, 20 situations), Wharton's Descent of Man (probe/wharton, 262 scenes, 16 groups) and Jacobs's The Lady of the Barge (probe/jacobs, 157 scenes, 19 groups). Each has its own README and a readable page. No inference was used for any of them. Catalogue. probe/catalogue maps every hand group in the three references onto 36 situation entries, with an answer key per corpus and one recurrence rule applied to all three. classify.py assigns a scene one entry or none, leave-one-corpus- out; score.py checks it against the key, with a self-test on random labels. Why clustering is out: hand-written predicaments, embedded and clustered exactly as the model's were, agree with the hand grouping at ARI 0.05 — no better than the 14B text's 0.07. Better rewriting cannot rescue it. Embeddings cannot even shortlist: the hand label is the nearest entry 13% of the time and in the top 8 half the time. The classification runs are not here. The dev and test runs are pre-registered in probe/catalogue/README.md with the bar set beforehand, and are blocked on the inference host, whose GPU has fallen off the PCIe bus three times. The 30-scene partial output in out/ is not a result. Review page. The situation review is now a browser page rather than JSON edited by hand (review_page.py, review_page_logic.cjs with Node tests, format schema v2). It has never been rendered in a real browser. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014BygvsUXV9eU6oHkTCkKZ1
This commit is contained in:
co-authored by
Claude Opus 5
parent
fa3769d0fe
commit
e5617b86ba
@@ -0,0 +1,25 @@
|
||||
[
|
||||
{"chunk":816,"story":94,"summary":"A miner back from the far north with a fortune sees a young woman in the homeward crowd and resolves at once to follow her.","predicament":"Someone with money and little time decides instantly on a stranger they have glimpsed."},
|
||||
{"chunk":817,"story":94,"summary":"She catches her ferry, exhausted after a late dance and a trying day, and he gains the deck beside her.","predicament":"Someone tired and discouraged is approached at their weakest moment."},
|
||||
{"chunk":818,"story":94,"summary":"He speaks; she takes him for a masher, then, seeing his sincerity, lets him sit and teases him about his name.","predicament":"Someone used to unwanted advances cautiously allows a sincere one."},
|
||||
{"chunk":819,"story":94,"summary":"He earnestly describes the gold dust he has brought south; she mistakes his meaning, grows bored, and falls asleep on his shoulder.","predicament":"Someone pours out their heart to a person who cannot stay awake to hear it."},
|
||||
{"chunk":820,"story":94,"summary":"He asks whether she would rather a husband had money or love; she cries 'Cash!' in her sleep, and wakes to find him gone.","predicament":"Someone loses a sincere suitor through a word spoken without meaning it."},
|
||||
{"chunk":821,"story":95,"summary":"A ten-dollar note tells its own history, noting that the cultured seldom hold it long enough to learn its manners.","predicament":"Something passed from hand to hand has seen every kind of owner."},
|
||||
{"chunk":822,"story":95,"summary":"In a butcher's purse it meets a limp five-dollar note that spent the day in a shop girl's stocking in the heat.","predicament":"Money earned by hard work arrives worn, while other money never touches it."},
|
||||
{"chunk":823,"story":95,"summary":"It circulates fast through bars, a peddler's roll, a policeman's pocket and a gambling back room, and delights in it.","predicament":"Someone's fortune passes through every sort of hand, clean and dirty alike."},
|
||||
{"chunk":824,"story":95,"summary":"A gambling-house owner rolls it into a wad for a night on the town, and a gold note says he will spend lavishly.","predicament":"Someone about to spend money freely has been refused the one use they wanted for it."},
|
||||
{"chunk":825,"story":95,"summary":"The gambler offered a church fifty thousand and was refused as tainted, and now half the town follows him to drink it.","predicament":"Someone's offered gift is refused as dirty, and others gladly take it instead."},
|
||||
{"chunk":826,"story":95,"summary":"At a cafe he asks whether the refused money will buy drink; at one o'clock a ragged woman comes in and stares at it.","predicament":"Someone desperate appears silently before a person who has money to spare."},
|
||||
{"chunk":827,"story":95,"summary":"He bows and hands her the note; once a Treasury clerk who counted such notes, she thanks him and runs to a bakery.","predicament":"Someone who fell from a respectable post accepts help from a person society scorns."},
|
||||
{"chunk":828,"story":95,"summary":"A dollar from her change reports she fed a sick, half-starved child and prayed; the ten says there is no such thing as tainted money.","predicament":"Someone's help does its good regardless of where the means came from."},
|
||||
{"chunk":829,"story":96,"summary":"A young woman's father, an old cutter, dies after an accident, leaving her a dollar and his employer's letter offering help.","predicament":"Someone left alone with almost nothing holds a promise of help from a powerful stranger."},
|
||||
{"chunk":830,"story":96,"summary":"Evicted, she rides downtown with her dollar and the letter to seek the firm, which has moved to an address she lacks.","predicament":"Someone sets out alone to find the one person who promised help."},
|
||||
{"chunk":831,"story":96,"summary":"A kindly western rancher looking for a wife approaches her honestly; a policeman seizes him as a masher and drags him away.","predicament":"Someone is shielded from a genuine chance by a guardian who assumes the worst."},
|
||||
{"chunk":832,"story":96,"summary":"Preferring to support herself, she is offered a companion's post by a kind lady, until an association agent forbids her to accept.","predicament":"Someone wanting to earn their own living is stopped by people claiming to protect them."},
|
||||
{"chunk":833,"story":96,"summary":"She takes a cashier's place in a candy store, and a temperance crusader warns her that the sweets contain brandy.","predicament":"Someone takes honest work and is told it will damn others."},
|
||||
{"chunk":834,"story":96,"summary":"She gives up the job at the crusader's urging, who drives off in her carriage; next she tries sewing theatrical costumes.","predicament":"Someone gives up work at a reformer's word and is left with nothing."},
|
||||
{"chunk":835,"story":96,"summary":"A clergyman forbids her to sew costumes for the stage and promises only that the Lord will provide.","predicament":"Someone in need is advised by a moralist who offers nothing practical."},
|
||||
{"chunk":836,"story":96,"summary":"Striking girls warn her off a factory job; at last she finds her father's employer, who welcomes her warmly and offers a place.","predicament":"Someone blocked from every honest job at last reaches the one person who promised help."},
|
||||
{"chunk":837,"story":96,"summary":"He has her try on furs and a fine hat as a model, then telephones to reserve his usual private table for a new companion.","predicament":"Someone driven by every protector to the promised benefactor finds a predator waiting."},
|
||||
{"chunk":838,"story":96,"summary":"The narrator blames the reformers and associations, compassionate in heart but reverent of money, for losing such girls every day.","predicament":"Someone well-meaning closes every honest door and leaves the poor only the dangerous one."}
|
||||
]
|
||||
Reference in New Issue
Block a user