log_adjacency
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
log_adjacency has 20 facts recorded in Dontopedia across 8 references, with 1 live disagreement.
20 facts·14 predicates·8 sources·1 in dispute
Mostly:rdf:type(4), updated by(2), gets gradient through(1)
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.
appliedToApplied to(1)
- Softmax Operation
ex:softmax-operation
includesIncludes(1)
- Oscillator Params
ex:oscillator-params
updatesParameterUpdates Parameter(1)
- Oscillator Param Update
ex:oscillator-param-update
usesSoftmaxParameterizationForUses Softmax Parameterization for(1)
- Root Fix
ex:root-fix
Other facts (18)
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.
18 facts
| Predicate | Value | Ref |
|---|---|---|
| Rdf:type | Parameter Tensor | [4] |
| Rdf:type | Parameter | [5] |
| Rdf:type | Model Parameter | [7] |
| Rdf:type | Model Parameter | [8] |
| Updated by | Kuramoto Energy Gradient | [1] |
| Updated by | Kuramoto Energy Gradient | [6] |
| Gets Gradient Through | Same Chain | [2] |
| Is Structural Constant | true | [3] |
| Not Learned | true | [3] |
| Is Static Tensor | true | [4] |
| Tensor Dimensions | G, G | [4] |
| Independent of | T | [4] |
| Has Softmax Applied | true | [4] |
| Has Value | 0 | [4] |
| Softmax Result | 1/G | [4] |
| Receives Gradient Through | Same Chain | [5] |
| Gradient Requirement | capturing spectra before and after sync | [5] |
| Has Status in Claude Md | Structural Constant | [8] |
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.
—
updatedByblah/watt-activation/part-193
ex:kuramoto-energy-gradient
—
getsGradientThroughblah/watt-activation/part-189
ex:same-chain
—
isStructuralConstantblah/watt-activation/part-424
true
—
notLearnedblah/watt-activation/part-424
true
—
labelblah/watt-activation/138
log_adjacency
—
typeblah/watt-activation/138
ex:ParameterTensor
—
isStaticTensorblah/watt-activation/138
true
—
tensorDimensionsblah/watt-activation/138
G, G
—
independentOfblah/watt-activation/138
ex:T
—
hasSoftmaxAppliedblah/watt-activation/138
true
—
hasValueblah/watt-activation/138
0
—
softmaxResultblah/watt-activation/138
1/G
—
typeblah/watt-activation/189
ex:Parameter
—
receivesGradientThroughblah/watt-activation/189
ex:same-chain
—
gradientRequirementblah/watt-activation/189
capturing spectra before and after sync
—
updatedByblah/watt-activation/193
ex:kuramoto-energy-gradient
—
typeblah/watt-activation/198
ex:ModelParameter
—
typeblah/watt-activation/422
ex:ModelParameter
—
labelblah/watt-activation/422
log_adjacency
—
hasStatusInClaudeMdblah/watt-activation/422
ex:structural-constant
References (8)
8 references
ctx:discord/blah/watt-activation/part-193ctx:discord/blah/watt-activation/part-189ctx:discord/blah/watt-activation/part-424ctx:discord/blah/watt-activation/138- full textwatt-activation-138text/plain3 KB
doc:agent/watt-activation-138/9bd08756-4bea-4055-bcfd-5d3ed62434a1Show excerpt
[2026-03-09 06:23] xenonfun: ``` ⏺ No — softmax is on self.log_adjacency which is a static (G, G) parameter tensor, completely independent of T. It runs once per forward pass in O(G²) = O(64). The sequence-length work is entirely in _gate…
ctx:discord/blah/watt-activation/189- full textwatt-activation-189text/plain2 KB
doc:agent/watt-activation-189/ee6e7700-8f8f-458c-bd97-cd00204ffe29Show excerpt
[2026-03-10 03:42] xenonfun: ``` What the fix looks like: Coupling κ_g is a scalar per group. Its gradient through the sync step is tractable: at first order, Δcoupling_g ∝ -(∂loss/∂spectra_synced) · (mean_spec_g - spectra_g) — the reado…
ctx:discord/blah/watt-activation/193- full textwatt-activation-193text/plain3 KB
doc:agent/watt-activation-193/b982ee37-c42f-49ed-bcc9-0f5b6259a2c9Show excerpt
[2026-03-10 04:26] lisamegawatts: if its now unfrozen, try the energy loss one [2026-03-10 04:26] xenonfun: ``` Root cause: The loss-gradient-derived coupling update is structurally anti-synchronizing. Coupling should be driven by Kuramoto …
ctx:discord/blah/watt-activation/198- full textwatt-activation-198text/plain2 KB
doc:agent/watt-activation-198/ea2dbbd7-7d1c-4ba2-9281-56322a048288Show excerpt
[2026-03-10 06:11] xenonfun: ``` ⏺ --- Summary #1 — features tensor restored One line in lohe_ffn.py. The optimizer already had the branch ready (if 'features' in metrics:). What activates now: - W_out[:, 32..92] (weighted harmon…
ctx:discord/blah/watt-activation/422- full textwatt-activation-422text/plain3 KB
doc:agent/watt-activation-422/a495a8d5-2cf4-41be-a842-8a6160f7c013Show excerpt
[2026-03-20 01:24] xenonfun: Flaw 4: Optimizer micromanaging — INTENTIONALLY ADDRESSED (different fix) This one is nuanced. CLAUDE.md says: "Coupling gradient updates REMOVED: K and log_adjacency are structural constants, not learned." T…
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.