coupling K
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
coupling K has 44 facts recorded in Dontopedia across 14 references, with 6 live disagreements.
Mostly:rdf:type(6), is structural(3), is structural constant(2)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (4)
Other subjects in dontopedia point AT this entity as a value. These are inverse relationships — e.g. "X motherOf this subject" — and answer questions the forward facts can't. Grouped by predicate.
causeDivergenceCause Divergence(1)
- Packed Batches
ex:packed-batches
describesMetricDescribes Metric(1)
- Dashboard Panel Row 3
ex:dashboard-panel-row-3
hasMemberHas Member(1)
- Two Matrices
ex:two-matrices
mentionsMentions(1)
- Log Entry 2026 03 12 08 11
ex:log-entry-2026-03-12-08-11
Other facts (41)
The long tail: predicates that appear too rarely to warrant their own section. Filter or scroll to find a specific one. Each row links to its source.
| Predicate | Value | Ref |
|---|---|---|
| Rdf:type | Structural Constant | [8] |
| Rdf:type | Metric | [10] |
| Rdf:type | Visualization Panel | [11] |
| Rdf:type | Visualization | [12] |
| Rdf:type | Visualization | [13] |
| Rdf:type | Hyperparameter | [14] |
| Is Structural | True | [6] |
| Is Structural | true | [9] |
| Is Structural | true | [11] |
| Is Structural Constant | True | [2] |
| Is Structural Constant | true | [8] |
| Has Value | 0.176 | [9] |
| Has Value | 0.177 | [14] |
| Is Learned | false | [9] |
| Is Learned | false | [11] |
| Approximate Value | 0.17 | [10] |
| Approximate Value | 0.18 | [10] |
| Differs From | Gate G | [1] |
| Computed Via | Coupling Computer | [1] |
| Shapes | Dynamics | [1] |
| Asks | who should influence whom's phase? | [1] |
| Not Updated | True | [2] |
| Locks to Value | ~0.176 | [3] |
| Confirms Fix | Anti Pattern | [3] |
| Barely Moves | True | [3] |
| Is Structural Not Learned | True | [3] |
| Is Middle Right Panel | true | [4] |
| Values Diverge Faster Than Before | true | [4] |
| Is Stable Across Blocks | ~0.16-0.18 | [5] |
| Shows No Runaway | true | [5] |
| Matches Expectation | True | [6] |
| Is Not Learned | True | [6] |
| Computed by | Coupling Computer | [7] |
| Function | Shapes Dynamics | [7] |
| Query | Influence Question | [7] |
| Is Not Updated | true | [8] |
| Applies to | All Blocks | [9] |
| Has Volatility | Low | [9] |
| Location in Layout | middle-right | [12] |
| Shows Feature | clear-divergence | [12] |
| Has Location | Middle Right Position | [13] |
Timeline
Timeline axis is valid_time — when each source says the fact was true in the world, not when Dontopedia learned about it. Retracted rows are kept for provenance; coloured stripes indicate the context kind.
References (14)
ctx:discord/blah/watt-activation/part-65ctx:discord/blah/watt-activation/part-210ctx:discord/blah/watt-activation/part-233ctx:discord/blah/watt-activation/part-250ctx:discord/blah/watt-activation/part-251ctx:discord/blah/watt-activation/part-247ctx:discord/blah/watt-activation/65- full textwatt-activation-65text/plain3 KB
doc:agent/watt-activation-65/3eb67599-95d0-41e7-8dbb-3ef5adce2574Show excerpt
[2026-03-07 15:47] xenonfun: ``` What the simplification would lose The "near-zero-cost drop-in" claim is mostly right for a single-layer, single-step model. But there are structural features the Kuramoto architecture provides that a n…
ctx:discord/blah/watt-activation/209- full textwatt-activation-209text/plain2 KB
doc:agent/watt-activation-209/d7eabcf9-d506-432f-9a2c-b25cf5ef8ccdShow excerpt
[2026-03-11 03:51] xenonfun: ```# coupling (K) and adjacency are structural constants — not updated. 598 + # v2: update harmonic_coeffs (G×H learned frequency weights) 599 + # v3: update mode_amplitudes (k scalars,…
ctx:discord/blah/watt-activation/232- full textwatt-activation-232text/plain3 KB
doc:agent/watt-activation-232/3c01e7f2-4764-4305-aaa5-72adde3c0ccbShow excerpt
[2026-03-11 20:31] xenonfun: ## SpectralAttention vs. AnchorKAN (files: Screenshot_2026-03-11_at_4.31.27_PM.png) [2026-03-11 20:39] xenonfun: ``` ⏺ Here's what each panel is showing, with the interesting bits: --- Row 1 — Loss + Global…
ctx:discord/blah/watt-activation/235- full textwatt-activation-235text/plain2 KB
doc:agent/watt-activation-235/2b477fba-6b21-4139-a3e7-6c48c5dad99fShow excerpt
[2026-03-11 22:42] xenonfun: ⏺ Done. Pushed 34b7246. Summary of what landed: - Results: anchor_kan best ppl 155.6 / 52K tok/s, spectral 116.4 / 43.7K tok/s - parallel prefill: 120× faster for spectral, auto-fallback for anchor_kan - …
ctx:discord/blah/watt-activation/246- full textwatt-activation-246text/plain3 KB
doc:agent/watt-activation-246/e98a7e00-4e3f-4e35-bccb-faff629fa820Show excerpt
[2026-03-12 07:06] xenonfun: ```[03:04:05] step 2000/12432 16.1% txt=4.411(82ppl) img=0.000 aud=2.025 r=0.375 lr=4.79e-05 150ms 69,028tok/s [T:1000 I:0 A:1000] eta=26min VAL step 2000 loss=4.4035 ppl=81.7 * best [03:04:29] …
ctx:discord/blah/watt-activation/248- full textwatt-activation-248text/plain2 KB
doc:agent/watt-activation-248/aaa8be70-5afb-4a65-b9f4-713720917264Show excerpt
[2026-03-12 08:11] xenonfun: ⏺ That's v3 epoch 1 (the full COCO+audio+text run). Notable evolution compared to earlier: r heatmap (top-left): Same block stratification — B0-B1 hot (r~0.5), B4-B5 cool (r~0.37). Stable throughout, no phase…
ctx:discord/blah/watt-activation/249- full textwatt-activation-249text/plain2 KB
doc:agent/watt-activation-249/8436e2be-67b0-44b5-9c54-97c19c68045dShow excerpt
[2026-03-12 08:24] xenonfun: ⏺ Early packed phase data — only ~550 steps in but already showing interesting dynamics: r heatmap (top-left): Much more visible step-to-step fluctuation than the unpacked runs. The vertical striping pattern …
ctx:discord/blah/watt-activation/351- full textwatt-activation-351text/plain3 KB
doc:agent/watt-activation-351/a1067b53-0e70-4497-b6fc-75de08ee45faShow excerpt
[2026-03-16 22:01] xenonfun: still working on unshackling our dynamics from locking in, now that we removed most of the shackles claude put on it with non-manifold preserving operations for most of it. finally getting some movement, tho thi…
See also
Keep researching
Missing something or suspicious of what's here? Kick off a research session — a Claude agent will investigate, cite its sources, and file new facts into a dedicated context you can review before accepting into the shared view.