Pairwise L512
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Pairwise L512 has 17 facts recorded in Dontopedia across 3 references, with 1 live disagreement.
Mostly:has iterations per second(3), has r value(2), has best ppl(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedHas R Valuein disputehasRValue
Has Iterations Per SecondhasIterationsPerSecond
Has Best PplhasBestPpl
- 19.3[2]all time · Part 53
Has Effective Anchor UtilizationhasEffectiveAnchorUtilization
- -[2]all time · Part 53
Has Final PplhasFinalPpl
- 70.1[2]all time · Part 53
Has Training Time in SecondshasTrainingTimeInSeconds
- 231.2[2]all time · Part 53
Is Instance ofisInstanceOf
- Pairwise Method[2]all time · Part 53
Uses Sequence LengthusesSequenceLength
- 512[2]all time · Part 53
Had R Value on Old CorpushadRValueOnOldCorpus
- 0.86[1]all time · Part 60
Has Final PplhasFinalPPL
- 80.1[1]all time · Part 60
Has R Value on Cleaned CorpushasRValueOnCleanedCorpus
- 0.587[1]all time · Part 60
R Dropped FromrDroppedFrom
- 0.86[1]all time · Part 60
Inbound mentions (4)
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.
hasHigherBestPplThanHas Higher Best Ppl Than(1)
- Anchor V3 L512
ex:anchor-v3-l512
hasLowerFinalPplThanHas Lower Final Ppl Than(1)
- Anchor V3 L512
ex:anchor-v3-l512
hasNearlyIdenticalThroughputToHas Nearly Identical Throughput to(1)
- Anchor V3 L512
ex:anchor-v3-l512
narrowsBetweenNarrows Between(1)
- Throughput Gap L512
ex:throughput-gap-l512
Other facts (2)
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.
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 (3)
- custom
ctx:discord/blah/watt-activation/part-60 - custom
ctx:discord/blah/watt-activation/part-53 - custom
ctx:discord/blah/watt-activation/62- full textwatt-activation-62text/plain2 KB
doc:agent/watt-activation-62/a54e16c9-84e0-4825-8097-9ebef4dc861dShow excerpt
[2026-03-07 14:44] xenonfun: L512/L1024/L2048 data [2026-03-07 14:50] xenonfun: (files: phase_diagram_r_vs_ppl.png, training_trajectories_r_vs_ppl.png, scaling_r_and_ppl_vs_seqlen.png) [2026-03-07 14:51] xenonfun: ``` +### Per-configuratio…
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.