ConstellationDecoder
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
ConstellationDecoder has 52 facts recorded in Dontopedia across 12 references, with 7 live disagreements.
Mostly:has parameter count(4), rdf:type(4), dominates parameter count(2)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (11)
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.
involvesEntityInvolves Entity(2)
- Hypothesis Capacity Allocation
ex:hypothesis-capacity-allocation - Hypothesis Constellation Decoder Specialization
ex:hypothesis-constellation-decoder-specialization
conditionalOnDecoderConditional on Decoder(1)
- K Probes Parameter
ex:k-probes-parameter
hasClassHas Class(1)
- Geometric Primitives
ex:geometric-primitives
lacksLacks(1)
- Model Variant B Cals Only
ex:model-variant-b-cals-only
mattersOnlyWithMatters Only With(1)
- K Probes Parameter
ex:k-probes-parameter
mentionsComponentMentions Component(1)
- Message 01 44 Stats
ex:message-01-44-stats
needsBuildingNeeds Building(1)
- Resonant Wire
ex:resonant-wire
proposedAsDecoderProposed As Decoder(1)
- Encoding Table
ex:encoding-table
relevanceConditionRelevance Condition(1)
- Parameter K Probes
ex:parameter-k-probes
usesUses(1)
- Resonant Wire
ex:resonant-wire
Other facts (49)
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 |
|---|---|---|
| Has Parameter Count | 32000 | [4] |
| Has Parameter Count | 32000 | [8] |
| Has Parameter Count | 32000 | [9] |
| Has Parameter Count | 23808 | [12] |
| Rdf:type | Design Principle | [7] |
| Rdf:type | Model Component | [8] |
| Rdf:type | Mode or Component | [10] |
| Rdf:type | Model Component | [12] |
| Dominates Parameter Count | true | [2] |
| Dominates Parameter Count | Model | [4] |
| Present in Variants | Model Variant a Global | [6] |
| Present in Variants | Model Variant C Coherence Gated | [6] |
| Constitutes Percentage of | 94 | [8] |
| Constitutes Percentage of | 84 | [12] |
| Part of Model | Model Variant a | [11] |
| Part of Model | Model Variant C | [11] |
| Is Most Interesting Part | {} | [1] |
| Has One Per | Byte | [1] |
| Avoids | Linear D to 256 | [1] |
| Avoids Flat Projection | {} | [1] |
| Is Manifold Aware | {} | [1] |
| Replaces | Linear Readout | [1] |
| Computes | Hermitian Similarity to Byte Templates | [1] |
| Computes Hermitian Similarity Between | Output State and 256 Learned Constellation Points | [1] |
| Has Dimensions | 256 × 125 | [2] |
| Param Dominant Component | true | [2] |
| Constitutes Percentage of Model | 94% | [2] |
| Has Memory Complexity | O(T × 256) | [3] |
| Implies Vocab Size256 | null | [3] |
| Has Compute Complexity | O(256 × readout_dim) | [3] |
| Constitutes94 Percent of | Model Parameters | [4] |
| Exhibits Design Smell | Design Smell | [4] |
| Dominates Total Params | 0.84 | [5] |
| Has Parameters | 23808 | [5] |
| Uses Readout Dim | 93 | [5] |
| Produces Higher Entropy Outputs on | Synthetic Patterns | [6] |
| At Expense of | Synthetic Patterns | [6] |
| Part of | Delta Memory Variants | [6] |
| Allocates Capacity to | Natural Text Distribution | [6] |
| Specialized for | Lm Text | [6] |
| Uses Method | Hermitian Similarity | [7] |
| Excludes Component | Linear D 256 | [7] |
| Characterized As | most interesting part | [7] |
| Computation Method | compute Hermitian similarity between the output state and 256 learned constellation points | [7] |
| Property | manifold-aware rather than a flat projection | [7] |
| Constitutes Percentage of Model Parameters | 94 | [9] |
| Described As | a design smell | [9] |
| Contrasts | Symmetric Mode | [10] |
| Has Readout Dimension | 256 × 93 | [12] |
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 (12)
ctx:discord/blah/watt-activation/part-384ctx:discord/blah/watt-activation/part-385ctx:discord/blah/watt-activation/part-401ctx:discord/blah/watt-activation/part-402ctx:discord/blah/watt-activation/part-607ctx:discord/blah/watt-activation/part-606ctx:discord/blah/watt-activation/382- full textwatt-activation-382text/plain3 KB
doc:agent/watt-activation-382/7400f8cd-0f1a-470c-9260-c96918758eaeShow excerpt
[2026-03-19 01:37] xenonfun: ⏺ This is gold. Let me check the sweep results while I synthesize the architecture idea: ``` Here's the stripped-down principled design, pulling from the harmonic-gpt findings: "Resonant Wire" — Helmholtz+Cli…
ctx:discord/blah/watt-activation/383ctx:discord/blah/watt-activation/400- full textwatt-activation-400text/plain2 KB
doc:agent/watt-activation-400/bfd3ae1a-a87b-4ef9-bd0b-548fd78cc0cbShow excerpt
[2026-03-19 05:11] xenonfun: ⏺ The ConstellationDecoder is 94% of the model's parameters (32K of 27K dynamics). That's a design smell. ``` The most elegant option: use the encoding table itself as the decoder. The BPSK table maps each byt…
ctx:discord/blah/watt-activation/406- full textwatt-activation-406text/plain1 KB
doc:agent/watt-activation-406/d69009ec-32be-45d6-b3d2-0b3143e59d42Show excerpt
[2026-03-19 06:36] xenonfun: ⏺ Here's what each knob does: G (groups) = 8 — How many independent oscillator groups. Each group reads a different bit of the byte. More groups = more parallel information channels, but more Givens coupling …
ctx:discord/blah/watt-activation/603- full textwatt-activation-603text/plain3 KB
doc:agent/watt-activation-603/7aa9fa83-a058-43b0-b700-073a69d3e610Show excerpt
[2026-04-10 04:25] xenonfun: ``` ⏺ Now we have real eval results on the trained models. Key observations: Results Summary (10K steps, 50% curriculum, trained model eval) ┌─────────────────┬────────────┬───────────────┬────────────────…
ctx:discord/blah/watt-activation/604- full textwatt-activation-604text/plain2 KB
doc:agent/watt-activation-604/5619f2b9-2e60-495c-a6e6-9fe7ca80a696Show excerpt
[2026-04-10 04:27] xenonfun: ``` From the training output logs, the models we've been testing have 28,525 total parameters: - Constellation (decoder): 23,808 params (256 × 93 readout dim) - Readout: 153 params (mode amplitudes, pro…
See also
- Byte
- Linear D to 256
- Linear Readout
- Hermitian Similarity to Byte Templates
- Output State and 256 Learned Constellation Points
- Model
- Model Parameters
- Design Smell
- Synthetic Patterns
- Delta Memory Variants
- Model Variant a Global
- Model Variant C Coherence Gated
- Natural Text Distribution
- Lm Text
- Design Principle
- Hermitian Similarity
- Linear D 256
- Model Component
- Mode or Component
- Symmetric Mode
- Model Variant a
- Model Variant C
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.