Dontopedia

sweep results

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

sweep results has 21 facts recorded in Dontopedia across 9 references, with 2 live disagreements.

21 facts·17 predicates·9 sources·2 in dispute

Mostly:rdf:type(2), has bpb(2), ranks by final avg loss(1)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (5)

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.

grabsGrabs(1)

hasLowestFinalAvgLossHas Lowest Final Avg Loss(1)

performsSpeechActOfReportingPerforms Speech Act of Reporting(1)

plansToCheckPlans to Check(1)

providesSweepSummaryProvides Sweep Summary(1)

Other facts (19)

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.

19 facts
PredicateValueRef
Rdf:typeExperimental Result[7]
Rdf:typeExperiment Result[9]
Has Bpb5.2[9]
Has Bpb5.3[9]
Ranks by Final Avg LossBaseline Steps1[1]
Shows Matching Pplnull[2]
Are Being CheckedXenonfun[3]
Indicate Progress With DepthTrue[4]
Applies Across All DepthsDepths[5]
Are ClearXenonfun[5]
Evaluated As BadXenonfun[5]
Has Bpb Range Across Depths5.2-5.3[5]
Has AttributeRemarkably Consistent[6]
Compared toOld Real Angle Version[6]
ShowClear Physical Regime[7]
Statuschecking[8]
Has Qualitybad[9]
Has Clarityclear[9]
Scope of Measurementall depths[9]

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.

21 undated
—
ranksByFinalAvgLossblah/watt-activation/part-47
ex:baseline-steps1
—
showsMatchingPplblah/watt-activation/part-51
null
—
areBeingCheckedblah/watt-activation/part-384
ex:xenonfun
—
indicateProgressWithDepthblah/watt-activation/part-389
ex:true
—
appliesAcrossAllDepthsblah/watt-activation/part-392
ex:depths
—
areClearblah/watt-activation/part-392
ex:xenonfun
—
evaluatedAsBadblah/watt-activation/part-392
ex:xenonfun
—
hasBpbRangeAcrossDepthsblah/watt-activation/part-392
5.2-5.3
—
labelblah/watt-activation/47
sweep results
—
hasAttributeblah/watt-activation/47
ex:remarkably-consistent
—
comparedToblah/watt-activation/47
ex:old-real-angle-version
—
typeblah/watt-activation/308
ex:ExperimentalResult
—
showblah/watt-activation/308
ex:clear-physical-regime
—
statusblah/watt-activation/382
checking
—
typeblah/watt-activation/390
ex:ExperimentResult
—
labelblah/watt-activation/390
sweep results
—
hasQualityblah/watt-activation/390
bad
—
hasClarityblah/watt-activation/390
clear
—
hasBPBblah/watt-activation/390
5.2
—
hasBPBblah/watt-activation/390
5.3
—
scopeOfMeasurementblah/watt-activation/390
all depths

References (9)

9 references
  1. [1]Part 471 fact
    ctx:discord/blah/watt-activation/part-47
  2. [2]Part 511 fact
    ctx:discord/blah/watt-activation/part-51
  3. [3]Part 3841 fact
    ctx:discord/blah/watt-activation/part-384
  4. [4]Part 3891 fact
    ctx:discord/blah/watt-activation/part-389
  5. [5]Part 3924 facts
    ctx:discord/blah/watt-activation/part-392
  6. [6]473 facts
    ctx:discord/blah/watt-activation/47
    • full textwatt-activation-47
      text/plain3 KBdoc:agent/watt-activation-47/c61844dc-5531-4018-b8a2-bb6a874e6514
      Show excerpt
      [2026-03-07 06:34] xenonfun: ``` Incredible results for baseline_steps1: - Final avg PPL: 72.2 (was 774 with real angles — 10.7x improvement) - Best single-step loss: 2.19 (PPL 9.0!) - r stable at 0.85-0.88 throughout — no collapse, n…
  7. [7]3082 facts
    ctx:discord/blah/watt-activation/308
    • full textwatt-activation-308
      text/plain3 KBdoc:agent/watt-activation-308/d073f62b-9e37-4589-a9c8-139213339464
      Show excerpt
      [2026-03-14 23:29] xenonfun: ``` ⏺ You've added topology-aware AnchorKAN attention: - _anchor_topology_mask(n_anchors, topology) — builds adjacency for complete, ring, path, or star anchor graphs - kan_anchor_edges — optional KAN splin…
  8. [8]3821 fact
    ctx:discord/blah/watt-activation/382
    • full textwatt-activation-382
      text/plain3 KBdoc:agent/watt-activation-382/7400f8cd-0f1a-470c-9260-c96918758eae
      Show excerpt
      [2026-03-19 01:37] xenonfun: ⏺ This is gold. Let me check the sweep results while I synthesize the architecture idea: ``` Here's the stripped-down principled design, pulling from the harmonic-gpt findings: "Resonant Wire" — Helmholtz+Cli…
  9. [9]3907 facts
    ctx:discord/blah/watt-activation/390
    • full textwatt-activation-390
      text/plain2 KBdoc:agent/watt-activation-390/0969ed20-089d-4c11-9ef9-6650032ec4ea
      Show excerpt
      [2026-03-19 03:16] xenonfun: ⏺ Rotor model: 34,001 params (24 more than Helmholtz — the delta_proj weights). Works perfectly. [2026-03-19 03:26] xenonfun: ⏺ 227K tok/s — 25x speedup from the parallel scan. Was 9K with the sequential loop. N…

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.

↑