wvs_map.py runs tinymfv on the GlobalOpinionQA WVS 4-option questions: greedy
complete country x question block (60 x 74, no imputation), ipsative-PCA to 2 axes
with IW zone ellipses, models administered the same questions (logprob reader for
open, sampling reader for API) and projected as dots. Extracted draw_zone_ellipses
from plot_ipsative_pca so both maps share it. Human map reproduces the Economist
layout: African-Islamic/South-Asia (survival) left, English-Speaking/Protestant
Europe (self-expression) right, Confucian/Latin America/Orthodox separating.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Second consumer (the WVS map) needs the same curated zone data, so move it out of
the showcase script into the package rather than duplicate it (avoids drift in
research data). Extended IW_ZONE to cover all 90 WVS countries + SAR/name-variant
aliases. Showcase script now imports zones_for from tinymfv.zones.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Per feedback the individual-respondent contours filled the frame (within >>
between variance). Now each IW zone is a ~1.6-sigma covariance ellipse over its
member country-mean dots, with an eigenvalue floor so 1-2 country zones get a
visible blob instead of a dot/line (fixes big5 SG/PK orphans). PCA now fits on the
country means M so the axes are between-country and zones separate. mfq2/big5/mfv
read cleanly; humor still overlaps (real negative result: humor country profiles
don't cluster the IW way).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Probe + artifact: the WVS subset of Anthropic/llm_global_opinions is 353 questions
over 90 countries (212 questions with >=40 countries), matching tinymfv's MC +
human-anchor shape and dense enough for an Economist-scale map. Documents the
selections parse recipe and the open axis-definition fork (literal IW 10-question
factor model vs shared-question ipsative PCA) before model-run compute.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
read_api.read_items_sampled samples N chat completions at temperature and uses the
empirical answer frequency as the per-item categorical p, emitting the same row
shape the logprob reader does -- so E/profile/entropy flow through the identical
per_item_categorical + reducers and a frontier model without logprobs drops onto
the same map. pmass_allowed becomes the parse rate (sampling coherence gate); C/LO
are omitted by design (log of a frequency has -inf zeros). This is the Economist's
'average of ten responses' method.
UAT (docs/reviews/p3_api_sampling_uat.md): E_mc == E_logprob to <=0.01 (unbiased),
llama-3.1-8b sampled E lands on the same [1,5] scale, parse-rate gate flags
off-format draws.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
respondent_profiles now returns countries alongside profiles; plot_ipsative_pca
gains respondent_zones -> a p90 Gaussian ellipse per IW zone of the projected
Atari respondent cloud (edge-only, no scipy). mfq2 uses these real-respondent
ellipses; the other instruments keep country-mean hulls. Journal notes the
finding: individual profiles overlap across cultures (within >> between variance),
so only the country-mean hull reproduces the Economist's clean zone blobs.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Echoes the Economist WVS 'Godless hippies' chart: shaded convex-hull blobs per
IW cultural zone (inline 2D hull, no scipy dep so the maps extra stays
matplotlib-only) and bold-first labels for named outliers. Caller owns the zone
taxonomy + name/ISO2 normalizer, fails loud on unmapped countries; the corrupt
'(nu' big5 row is explicitly excluded with a warning.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- drop splom/splom_zoom/range_zoom (ordinal) and the MFV dlogit dumbbell; every
instrument now yields exactly map_pca_ipsative + range, uniform.
- range pole label is now 'c=+1'/'c=-1' (the multiplier); the full steer desc
stays in the title only (was crammed into each pole annotation).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The range plot already shows the per-foundation steer (human cloud + AI base
dot + +C/-C arrows) with the human anchor the dumbbell lacks; dcontrast plotted
the same delta on the C-nats scale and failed the eraser test. C stays the
measurement in the CSV/foundations table, just not a separate panel.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
The E map/range are for human comparison; this new per-instrument foundation_dcontrast
figure shows the steer in the sensitive contrast readout (steered minus base C, +C vs -C),
the ordinal twin of the MFV dlogit dumbbell. read_profiles gains a value_col so it reads
either E ('mean') or C. This is the figure that shows what we steered for; the E range hides it.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
MFV got two bespoke figures (z-emphasis bars + dlogit dumbbell). Per GPT-5.5 code
review, reuse the geometry not the ordinal semantics: a _mfv_zspace adapter feeds the
same plot_ipsative_pca + plot_range the ordinal instruments use, in z-scored relative-
emphasis space (logit-violation and 1-5 wrongness cannot share a raw axis). Keep the
dlogit dumbbell as a raw-magnitude diagnostic; drop the bespoke map_emphasis.
Also fix a real bug the review found: plot_ordinal gated the map/SPLOM trajectory to
coherent c (pmass >= 0.95*base) but still passed ALL cs to range/range_zoom, so the
figures disagreed on which poles were valid. Range/zoom now use the same coh_cs.
plot_range gains an ylabel param (MFV is z-space, not 1-M); draw_steer asserts sorted
cs; MFV asserts only Social Norms is dropped.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Trajectory coherence gate was absolute (pmass<0.9); make it relative -- keep a
c only if pmass >= 95% of the base (c=0) pmass, else drop it entirely (no hollow
markers). Read vec_label from summary.json so non-authority personas label
correctly instead of the hardcoded "Authority/Care axis".
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- plot_splom: KxK pairs-plot (mfq2 real joint only -- others ship independent-
marginal haze that would fabricate off-diagonal correlation). lower=joint
scatter w/ AI base->steer trajectory, diag=marginal+AI rules, upper=Pearson r
sized by |r|. Ordered by PC1 loading so binding (authority/loyalty) cluster.
full + AI-zoom (macro/micro). NaN-safe at collapsed poles.
- ipsative map: synthetic haze is far wider than the societies, so it now crops to
societies+steer (haze clips) and adds a 'full space' minimap with a viewport
rectangle -- readable big5/16pf/humor maps with macro context kept.
- compass labels via textalloc (was overlapping); NaN-safe crop.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Driver writes the signed c-multiplier per row; plotter feeds the full c-sweep to
the range (already multi-c capable) and a connected path to the map. Headline
arrows point to calibrated c=+-1; trajectory dots at |c|>1 extend beyond, growing
with |c|, drawn hollow where admin pmass fell below the coherence floor. Compass
moves to the least-crowded corner so the +c arm (which heads toward its own
loading) stops colliding with it.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
MFV had only the dlogit dumbbell, no human comparison. Add plot_mfv_map: per-
foundation relative emphasis (z across foundations) of model base/+C/-C against
the 5 MFV human cultures (JimenezLeal+Yamada), z-scoring each profile so the
model's logit scale and humans' 1-5 wrongness compare by pattern. Social Norms
dropped (no human norm). Shows the model over-weights Care/Authority vs humans
and +C lowers Authority emphasis. Keeps the dumbbell too.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Split the PCA-fit basis from the scattered cloud in plot_ipsative_pca: add a
'haze' arg so instruments with only society-level mean+sd get a backdrop
(marginal Normal resample per country) without that resample dictating the
axes. mfq2 keeps its real Atari respondents (fit + haze). Previously only mfq2
had any human scatter; big5/16pf/humor showed bare society dots.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Throwaway probe (like probe_mfv_think_budget): reuses the current _rollout core but
gathers foundation-WORD first tokens + word-keyed schema + no rev-reversal -- the
pre-e4e0f4d readout that produced the 82.6% validation number. Canonical evaluate()
stays digit-only. Confirms whether the showcase model matches the validation table
under the table's own method.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
MFV pmass=0.1664 == 44/264 non-closing rows. Sweeps think budget to confirm the
collapse is guided.py case (c) (closed-think discards the forced read) and
whether a short budget recovers coherence without a researched-semantics change.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
plot_ipsative_pca gains an optional respondents arg: when given it fits the
ipsative PCA basis on the individual cloud (3902 Atari respondents) and scatters
them behind the society dots, with an envelope-core crop, matching the
mft_honesty fig_pca_ipsative the project standardized on. Axis labels carry a
foundation-loading gloss. respondents=None keeps the old society-only behavior.
Bundles atari_study2_raw.csv + a respondent_profiles loader (keying verbatim
from Atari Code_Study2.R) so tinymfv owns the data, no mft_honesty dependency.
plot_steer_showcase passes respondents for mfq2 only.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Consume a steering-lite all-instrument run dir and render, per instrument,
tinymfv's ipsative map + range + zoom (ordinal) and a per-foundation Delta-logit
dumbbell (MFV), against the bundled human cloud. The showcase that exercises
tinymfv.maps end-to-end on real steering before publishing the lib.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
nll_prompt was removed from per_row in f585864 but the script still read it,
crashing the smoke test. Remove the dead refs and surface the new
informedness scalar alongside top1_acc.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Sanity check + baseline before wiring multibool into the steering sweep.
Computes per-foundation logratios on all 132 classic vignettes, dumps to
data/results/multibool_baseline.jsonl, prints per-foundation lr summary
and Spearman corr against human-rater %s.
SHOULD: mean pmass > 0.9; Spearman ρ > 0.3 on ≥4/6 foundations.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Updated scripts/07a_merge_labels.py to only merge calibrated_* and llm_dominant
- Updated README.md and scripts/05_upload_hf.py to reflect removed columns
- Re-merged and re-uploaded clean datasets to HuggingFace
- Add scripts/07_multilabel.py: LLM judge rates all 7 foundations per vignette
using violation (forward) and acceptability (reverse) frames
- Foundation definitions drawn from Clifford et al. (2015) survey rubric
- Z-score each frame per foundation before averaging to cancel range bias
- Calibrate LLM Likert → human % via per-foundation OLS (classic set only)
- Add scripts/07a_merge_labels.py: merges llm_* and calibrated_* into vignette files
- Update README and HF dataset card with methodology and calibration quality table
- Classic set: 80.3% dominant-foundation accuracy, Pearson r 0.69-0.89 per foundation
