Dontopedia

GPU training path wired

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

GPU training path wired has 12 facts recorded in Dontopedia across 3 references, with 3 live disagreements.

12 facts·8 predicates·3 sources·3 in dispute

Mostly:rdf:type(2), involves method(2), pertains to component(1)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (1)

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.

containsCodeChangeContains Code Change(1)

Other facts (10)

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.

10 facts
PredicateValueRef
Rdf:typeProposed Code Change[1]
Rdf:typeCode Change[2]
Involves MethodResidentBackwardCache[3]
Involves Methodtrain_step_gpu_cached[3]
Pertains to ComponentSpectral Attention[1]
Describes TransformationQkv Fusion Transformation[1]
Affects FileRotor Attention V2 File[2]
Description IsAdded cv_all and v_norm_all to cache[2]
Improves Speed VsCPU[3]
Has Speedup Factor2.49[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.

labelblah/watt-activation/118
Fused QKV projection in SpectralAttention
typeblah/watt-activation/118
ex:ProposedCodeChange
pertainsToComponentblah/watt-activation/118
ex:SpectralAttention
describesTransformationblah/watt-activation/118
ex:qkv-fusion-transformation
typeblah/watt-activation/664
ex:CodeChange
affectsFileblah/watt-activation/664
ex:rotor-attention-v2-file
descriptionIsblah/watt-activation/664
Added cv_all and v_norm_all to cache
labelblah/watt-activation/681
GPU training path wired
involvesMethodblah/watt-activation/681
ResidentBackwardCache
involvesMethodblah/watt-activation/681
train_step_gpu_cached
improvesSpeedVsblah/watt-activation/681
CPU
hasSpeedupFactorblah/watt-activation/681
2.49

References (3)

3 references
  1. [1]1184 facts
    ctx:discord/blah/watt-activation/118
    • full textwatt-activation-118
      text/plain3 KBdoc:agent/watt-activation-118/ed79098d-1144-44f5-9941-e6b2b9c1caa7
      Show excerpt
      [2026-03-08 23:43] xenonfun: Code Changes (3 important patterns) 1. Fused QKV projection in SpectralAttention - Separate q_proj, k_proj, v_proj → single qkv_proj = Linear(d_model, 3 * d_model). One matmul instead of three. We should po
  2. [2]6643 facts
    ctx:discord/blah/watt-activation/664
  3. [3]6815 facts
    ctx:discord/blah/watt-activation/681
    • full textwatt-activation-681
      text/plain3 KBdoc:agent/watt-activation-681/376be3cc-9082-437f-bd74-a16443059d2a
      Show excerpt
      [2026-04-22 15:24] xenonfun: ``` Code changes this session: 1. GPU training path wired (train_step_gpu_cached via ResidentBackwardCache) — 2.49× vs CPU 2. Step-mode inference (UnifiedRotorBlock::step + UnifiedRotorLM::step + UnifiedRoto

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.