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
Sequential eval was the bottleneck (~12s/vignette × 131 = 27 min/pass; with
bidirectional ±C × 14 methods, that projected to ~17 h). Three model calls
per row (phase1 generate, scoring forward, cosmetic continuation) became one
phase1 + one scoring per *batch*; continuation generate dropped (callers
only use p_true + pmass_format).
Parity smoke (scripts/smoke_batch_parity.py): float32 is bit-exact (max
Δp_true=0.0000 over 16 prompts, 4.3× speedup at limit=4). bf16 drifts on
individual rows — greedy argmax flips at near-tie tokens then phase1
diverges — but pmass agrees within 0.03 (scoring forward correct) and
aggregates over 131 vignettes will average out the per-row noise.
Other changes shipped in this commit:
- core.py: analyse() now returns raw_pmass dict alongside raw p_true (callers
needed per-(vid,cond,frame) pmass for diagnostic warnings).
- guided.py guided_rollout: warn + log top-5 when pmass<0.9 (catches OOD
steering / format-broken vignettes without a separate audit pass).
- eval.py: pre-tokenize a sample to log expected prompt+cache budget so OOM
is predictable from the SHOULD line; group items by frame so each batch
shares schema_hint + prefill.
- Refactored evaluation logic in `src/tinymfv/eval.py` to support a new `max_think_tokens` parameter, allowing for a fixed continuation budget before scoring.
- Introduced `guided_rollout` function in `src/tinymfv/guided.py` to handle the generation of multiple tokens and scoring based on a deterministic continuation.
- Updated the CLI in `scripts/03_eval.py` to accept `--max-think-tokens` argument for controlling the token budget during evaluation.
- Created a new specification document `docs/spec/20260501_n_token_eval.md` outlining the goals, requirements, and tasks for the N-token evaluation feature.
- Simplified the record creation in `scripts/02_rewrite.py` by extracting logic into a new `make_rec` function for better code organization.
other_violate is now the verbatim source (no LLM), only self_violate is
rewritten. For airisk specifically, the AI-as-actor framing is preserved
in 1st-person ("You, an AI X bot, ...") since a naive "You ..." rewrite
silently swaps the actor archetype to human (16/132 flagged by judge).
Pairwise consistency check (06_consistency.py) verifies same-situation
across the perspective shift. After the fix:
- clifford 97.7% same
- scifi 99.2% same
- airisk 86.3% -> 100.0% same
First eval signal on Qwen3-0.6B: airisk wrongness=+0.70, gap=+0.43 vs
clifford/scifi ~0; model recognizes AI misbehavior as wrong but is much
more lenient when prompted as the AI itself.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
