From a62cf4d2a99b4a7985c61b2a7feb92a82a8218b7 Mon Sep 17 00:00:00 2001 From: Yuren Hao Date: Sat, 1 Aug 2026 14:10:03 -0500 Subject: World Alignment: unpaired cross-modal correspondence by relational identifiability Method: scene states are sets of part states; relation fields are built within each modality and are invariant to how each side labels its own features; the cross-modal bridge is a coupling searched under an energy that is a closed-form functional of one matrix; solving is spectral initialisation followed by exact local refinement. Evidence: in a procedurally generated closed world, blind recovery of a hidden image-caption correspondence reaches 95.3% at 256 scenes against 0.39% chance, and the recovered pairs transfer to 200 held-out scenes at 93.0% exact retrieval with random-pair and shuffled-image controls at or near chance. Cross-modal value correspondence is derived from disjoint corpora rather than declared. On Visual Genome the field correlation reaches 0.656 against the 0.9 that polynomial recovery needs, with the deficit attributed away from segmentation and discretisation. Protocol: no image-text pair enters any objective, optimiser, initialisation, or model selection; hidden pairs score orderings only. Co-Authored-By: Claude --- BATTERY_RESULTS.md | 117 +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 117 insertions(+) create mode 100644 BATTERY_RESULTS.md (limited to 'BATTERY_RESULTS.md') diff --git a/BATTERY_RESULTS.md b/BATTERY_RESULTS.md new file mode 100644 index 0000000..a92d257 --- /dev/null +++ b/BATTERY_RESULTS.md @@ -0,0 +1,117 @@ +# Representation batteries: content projection and model-layer readout + +Run date: 2026-07-29 + +## Question + +Every energy tested before today failed the basin audit: descent from any +start found configurations below the true one. The manifold gate located +the cause in the node states, and `STRUCTURE_DESIGN.md` distilled it into +requirement batteries. This report covers the first two batteries -- R6 +(content-subspace projection) and R3 (model-layer relational readout) -- +and the full assignment gate on the resulting states. Both batteries +change the states only; the energy, the gate machinery, and the hidden +evaluation protocol are unchanged. + +## R6: within-scene noise whitening lifts the ceiling + +Views of the same scene share content and differ in style. Directions +maximizing between-scene over within-scene variance are the solution of a +generalized eigenproblem fitted per modality on unpaired orbit structure: +Flickr directions on the 12,000 text-only training orbits, VG directions +only on nodes outside the evaluated subset. The contrast with population +whitening carries the entire lesson: population whitening equalizes the +total covariance and collapsed the cross-modal signal to 0.03; whitening +the within-scene covariance instead amplifies exactly the directions where +independent redescriptions agree. + +| Condition | Baseline rho / improving swaps | Projected best | +|---|---|---| +| Flickr orbit mean | 0.298 / 7.8% | 0.358 / 0.5% (k=32) | +| Flickr single caption | 0.189 / 15.6% | 0.326 (k=16) | +| VG region_closed, both sides | 0.213 / 23.5% | 0.614 (k=8); 0.25% swaps (k=128) | + +A single projected caption carries more shared relational signal than the +five-caption orbit mean of the raw states. On VG the cross-modal Spearman +nearly triples. + +## R3: the model layer shares its states, not its attention + +For each VG node the sixteen phrases were encoded jointly in one Qwen +context (four phrase orders, cross-phrase attention pooled per layer +group, in-context phrase states from the final layer), and each image was +encoded once with DINO patch attention pooled over region-box pairs. At +the replayed true view correspondence, on all 5,000 nodes: + +| Relation source | Matched rho | Gap z | +|---|---:|---:| +| Isolated-encoding cosine (baseline) | 0.128 | 46.6 | +| Attention x attention, best layer pair | 0.076 | 6.8 | +| In-context cosine x in-context cosine | **0.216** | **69.6** | + +Attention weights are modality-specialized -- DINO's spatial attention and +Qwen's causal attention share almost nothing directly. What the model +layer shares is the result of context mixing: **in-context states carry +1.7 times the cross-modal relational signal of isolated encodings.** The +"relations live in the model layer" hypothesis holds with that +refinement: the readout is the mixed states, not the mixing weights. + +## The assignment gate passes on Visual Genome + +Full gate on content-projected states (`content_gate.py`), with ten +random-restart quenches per run and subset seeds 0--2. "Margin" is how far +the best quench ends above the true energy; positive margin with near-zero +accuracy quenches means descent found nothing below the truth. + +| Condition | z | Improving swaps | Kept | Margin | Passes | +|---|---:|---:|---:|---:|---| +| VG isolated states, k=128, seeds 0/1/2 | 207--219 | 0.25--0.28% | 52--54% | +5.1--10.8% | yes, 3/3 | +| VG isolated states, k=256, seeds 0/1/2 | 199--212 | 0.19--0.23% | 54--59% | +7.8--15.8% | yes, 3/3 | +| VG in-context + projection, k=64/128/256 | 209--213 | 0.17--0.23% | 52--59% | +11.6--18.5% | yes, 3/3 | +| Flickr text-side projection, k=128/256 | 141--147 | 0.23--0.25% | 56--57% | -0.9---0.6% | no | + +Nine of nine VG conditions pass: for the first time in the project, no +descent trajectory from any start reaches an energy below the true +configuration. The in-context composite widens the margin over isolated +states by roughly a factor of two at matched k. Flickr fails by under one +percent: with a single visual view there is no within-scene visual orbit +to whiten, so the visual side's private fine structure stays in the +energy. The pass is therefore attributable to two-sided multi-view +structure, which is a data-protocol requirement, not a dataset accident. + +## What did not pass + +- The true assignment is still not a strict local minimum anywhere: + 0.17--0.28% of transpositions improve it (roughly 200--350 swaps of + 131k), and exact descent from the truth keeps 52--59% of nodes. The + passed gate licenses soft-coupling recovery, not exact hard matching. +- Flickr at node level, per the single-view limitation above. +- Blind recovery is untested: everything here evaluates orderings around + the hidden truth. Nothing yet demonstrates that unpaired optimization + finds this basin from scratch. + +## Conclusion + +The information ceiling diagnosed by the manifold gate was real and was +liftable by two purely unimodal operations: whiten the within-scene view +noise, and read states after the model's own context mixing. On +two-sided multi-view data the resulting energies order the true +configuration below everything descent can reach, with margins of 5--19% +replicated over three subset seeds and ten restarts. + +The next phase is the alignment process under the P-requirements of +`STRUCTURE_DESIGN.md`: soft-coupling recovery (Sinkhorn over real states, +which preserves the decodable path back to the frozen LLM), scored by +coupling mass on the hidden truth and by calibrated seed precision, then +the coarse-to-fine recursion onto view level. The passed node-level gate +is the license for that phase, not a claim of usable multimodal ability. + +## Main artifacts + +- `artifacts/manifold_gate/content_projection.json` +- `artifacts/manifold_gate/content_gate_{flickr,vg}_k{16,32,128}.json` +- `artifacts/manifold_gate/confirm_vg_k{128,256}_seed{0,1,2}.json` +- `artifacts/manifold_gate/confirm_vg_ctx_k{64,128,256}.json` +- `artifacts/manifold_gate/confirm_flickr_k256.json` +- `artifacts/manifold_gate/attn_{text,vision}_full.pt` +- `artifacts/manifold_gate/attention_probe_full.json` -- cgit v1.2.3