Dontopedia

proj_in

From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)

proj_in has 26 facts recorded in Dontopedia across 12 references, with 1 live disagreement.

26 facts·18 predicates·12 sources·1 in dispute

Mostly:rdf:type(3), is binary selection sign problem(2), is naturally discrete(2)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (14)

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.

hasComponentHas Component(4)

usesProjectionUses Projection(2)

acceleratesAccelerates(1)

erasesMagnitudeErases Magnitude(1)

isRawProjInOutputIs Raw Proj in Output(1)

  • Hex:h

lockedByLocked by(1)

  • Rex:r

modifiesModifies(1)

stateOfState of(1)

usesUses(1)

usesOperationUses Operation(1)

Other facts (23)

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.

23 facts
PredicateValueRef
Rdf:typeOperation[10]
Rdf:typeModel Layer[11]
Rdf:typeNn.linear Layer[12]
Is Binary Selection Sign Problemtrue[5]
Is Binary Selection Sign Problemtrue[11]
Is Naturally Discretetrue[5]
Is Naturally Discretetrue[11]
Fuses Identically toVanilla Mlp Layers[6]
Fuses Identically toVanilla Mlp Layers[12]
Flows toNormalize[1]
Uses InitDefault Xavier[2]
Updated by Loss Gradienttrue[3]
Applied Before K Adjustmenttrue[4]
Locks R Too Earlytrue[4]
With Ternary Weightsselects input features into oscillator groups[5]
Is Instance ofNn Linear Layer[6]
From InputSame Input X[7]
ProjectsAll Groups[7]
UsesStructured Filterbank[7]
Updated byLoss Gradient[9]
Effectively LocksR[10]
EffectLocks R Too Early[10]
Has Ternary Weightstrue[11]

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.

26 undated
—
flowsToblah/random/part-39
ex:normalize
—
usesInitblah/watt-activation/part-184
ex:default-xavier
—
updatedByLossGradientblah/watt-activation/part-193
true
—
appliedBeforeKAdjustmentblah/watt-activation/part-206
true
—
locksRTooEarlyblah/watt-activation/part-206
true
—
withTernaryWeightsblah/watt-activation/part-277
selects input features into oscillator groups
—
isBinarySelectionSignProblemblah/watt-activation/part-277
true
—
isNaturallyDiscreteblah/watt-activation/part-277
true
—
isInstanceOfblah/watt-activation/part-437
ex:nn-linear-layer
—
fusesIdenticallyToblah/watt-activation/part-437
ex:vanilla-mlp-layers
—
fromInputblah/watt-activation/part-423
ex:same-input-x
—
projectsblah/watt-activation/part-423
ex:all-groups
—
usesblah/watt-activation/part-423
ex:structured-filterbank
—
labelblah/watt-activation/188
proj_in
—
updatedByblah/watt-activation/193
ex:loss-gradient
—
typeblah/watt-activation/205
ex:Operation
—
labelblah/watt-activation/205
proj_in
—
effectivelyLocksblah/watt-activation/205
ex:r
—
effectblah/watt-activation/205
ex:locks-r-too-early
—
typeblah/watt-activation/275
ex:ModelLayer
—
labelblah/watt-activation/275
proj_in
—
hasTernaryWeightsblah/watt-activation/275
true
—
isBinarySelectionSignProblemblah/watt-activation/275
true
—
isNaturallyDiscreteblah/watt-activation/275
true
—
typeblah/watt-activation/435
ex:nn.Linear-Layer
—
fusesIdenticallyToblah/watt-activation/435
ex:vanilla-mlp-layers

References (12)

12 references
  1. [1]Part 391 fact
    ctx:discord/blah/random/part-39
  2. [2]Part 1841 fact
    ctx:discord/blah/watt-activation/part-184
  3. [3]Part 1931 fact
    ctx:discord/blah/watt-activation/part-193
  4. [4]Part 2062 facts
    ctx:discord/blah/watt-activation/part-206
  5. [5]Part 2773 facts
    ctx:discord/blah/watt-activation/part-277
  6. [6]Part 4372 facts
    ctx:discord/blah/watt-activation/part-437
  7. [7]Part 4233 facts
    ctx:discord/blah/watt-activation/part-423
  8. [8]1881 fact
    ctx:discord/blah/watt-activation/188
    • full textwatt-activation-188
      text/plain3 KBdoc:agent/watt-activation-188/0b24c5f9-ca6d-47b7-9d97-98b6fac36e0c
      Show excerpt
      [2026-03-10 03:16] xenonfun: well I imagine data from working RotAdamW will be informative for it as to how to correct behavior / step issues in LoheOptimizer [2026-03-10 03:17] xenonfun: also that will be recorded [2026-03-10 03:38] xenonf…
  9. [9]1931 fact
    ctx:discord/blah/watt-activation/193
    • full textwatt-activation-193
      text/plain3 KBdoc:agent/watt-activation-193/b982ee37-c42f-49ed-bcc9-0f5b6259a2c9
      Show 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 …
  10. [10]2054 facts
    ctx:discord/blah/watt-activation/205
    • full textwatt-activation-205
      text/plain2 KBdoc:agent/watt-activation-205/9ef261de-33ef-4e77-a9ad-af07b253a5ab
      Show excerpt
      [2026-03-11 03:09] lisamegawatts: <@1438866165475708979> how would you explain to a claude that proposed this why it is wrong: ⏺ Running in mac-mini:smoketest-4. While that runs — the coupling gradient is still wrong because K_target = (d-r…
  11. [11]2755 facts
    ctx:discord/blah/watt-activation/275
    • full textwatt-activation-275
      text/plain3 KBdoc:agent/watt-activation-275/149bf24b-54fb-4412-b289-e2e03bccffe2
      Show excerpt
      [2026-03-13 20:46] xenonfun: ⏺ This tells a clear story: Current trained weights are NOT naturally ternary. Only ~46% of values are near {-1, 0, 1} after scaling (random Gaussian would give ~47%, so no better than chance). Relative err…
  12. [12]4352 facts
    ctx:discord/blah/watt-activation/435
    • full textwatt-activation-435
      text/plain2 KBdoc:agent/watt-activation-435/6e80af4f-2aed-449f-9f6a-4750597bfb8e
      Show excerpt
      [2026-03-20 06:59] xenonfun: ``` ⏺ You're right. The PR #7 results (p=0.005, d=2.57, 5/5 seeds) were validated with a specific fusion operator — _block_diagonal_transfer() on vanilla nn.Linear layers. The fusion level (how weights map be…

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.

↑