Dontopedia

kv_inp tensor

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

kv_inp tensor has 12 facts recorded in Dontopedia across 3 references, with 1 live disagreement.

12 facts·9 predicates·3 sources·1 in dispute

Mostly:has shape(3), has size(1), is memory bottleneck(1)

Maturity scale raw canonical shape-checked rule-derived certified

Other facts (11)

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.

11 facts
PredicateValueRef
Has Shape(B,T,32,768)[1]
Has Shape(16, 2048, 8, 4, 104)[2]
Has Shape(16, 2048, 8, 4, 104)[3]
Has Size416MB[2]
Is Memory BottleneckCumsum[2]
Rdf:typeTensor[3]
Has Memory Size416000000[3]
Has Data Typefloat32[3]
Identified Asmemory bottleneck[3]
Unavoidable forcumsum[3]
Dominatesmemory pressure[3]

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.

hasShapeblah/watt-activation/part-203
(B,T,32,768)
hasSizeblah/watt-activation/part-293
416MB
isMemoryBottleneckblah/watt-activation/part-293
ex:cumsum
hasShapeblah/watt-activation/part-293
(16, 2048, 8, 4, 104)
labelblah/watt-activation/291
kv_inp tensor
typeblah/watt-activation/291
ex:Tensor
hasMemorySizeblah/watt-activation/291
416000000
hasShapeblah/watt-activation/291
(16, 2048, 8, 4, 104)
hasDataTypeblah/watt-activation/291
float32
identifiedAsblah/watt-activation/291
memory bottleneck
unavoidableForblah/watt-activation/291
cumsum
dominatesblah/watt-activation/291
memory pressure

References (3)

3 references
  1. [1]Part 2031 fact
    ctx:discord/blah/watt-activation/part-203
  2. [2]Part 2933 facts
    ctx:discord/blah/watt-activation/part-293
  3. [3]2918 facts
    ctx:discord/blah/watt-activation/291
    • full textwatt-activation-291
      text/plain3 KBdoc:agent/watt-activation-291/883c968e-0b03-4e27-aca9-027dfc155696
      Show excerpt
      [2026-03-14 03:41] xenonfun: Why Keeping the Spherical Head Still Makes Sense (performace at 600K parm scale is effectively same as euclidian head) ``` Even if performance is the same, the spherical head is still the better design. Reasons

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.