Dontopedia

Standard Transformer

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

Standard Transformer has 5 facts recorded in Dontopedia across 3 references.

5 facts·5 predicates·3 sources

Mostly:is inefficient(1), wastes capacity(1), inference kv cache growth(1)

Maturity scale raw canonical shape-checked rule-derived certified

Other facts (5)

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.

5 facts
PredicateValueRef
Is InefficientEmbedding Table[1]
Wastes CapacityLearned Embedding Table[1]
Inference Kv Cache GrowthLinear With Context[2]
Rdf:typeArchitecture[3]
Has ComponentEmbedding Table[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.

5 undated
—
isInefficientblah/watt-activation/part-333
ex:embedding-table
—
wastesCapacityblah/watt-activation/part-333
ex:learned-embedding-table
—
inferenceKvCacheGrowthblah/watt-activation/126
ex:linear-with-context
—
typeblah/watt-activation/331
ex:Architecture
—
hasComponentblah/watt-activation/331
ex:embedding-table

References (3)

3 references
  1. [1]Part 3332 facts
    ctx:discord/blah/watt-activation/part-333
  2. [2]1261 fact
    ctx:discord/blah/watt-activation/126
    • full textwatt-activation-126
      text/plain3 KBdoc:agent/watt-activation-126/dddfc295-807c-4943-b01a-f4f0a977c17e
      Show excerpt
      [2026-03-09 04:03] xenonfun: ### What context count we do at this scale? ⏺ From the measurements we have, memory scales roughly linearly with total tokens in the batch: - BS=4, seq=1024 → 4,096 tokens → ~40 GB - BS=8, seq=1024 → 8,192 …
  3. [3]3312 facts
    ctx:discord/blah/watt-activation/331
    • full textwatt-activation-331
      text/plain3 KBdoc:agent/watt-activation-331/171bcb73-6b34-47a5-8430-e89c28ce4ad9
      Show excerpt
      [2026-03-15 06:04] xenonfun: ``` Correct — nobody is doing this. Based on the literature review: The closest anyone gets is RoPE, which is mathematically PSK but nobody in the ML community frames it that way or designed it from that perspec…

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.

↑