Dontopedia

professional_benchmark

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

professional_benchmark has 38 facts recorded in Dontopedia across 8 references, with 3 live disagreements.

38 facts·32 predicates·8 sources·3 in dispute

Mostly:row count(2), rdf:type(2), raw doremi weight(2)

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.

causedExcessiveExposureCaused Excessive Exposure(1)

causesLowSamplingWeightCauses Low Sampling Weight(1)

decreasedWeightForDecreased Weight for(1)

observesGovernorEffectObserves Governor Effect(1)

usesDatasetUses Dataset(1)

Other facts (36)

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.

36 facts
PredicateValueRef
Row Count218[5]
Row Count218[6]
Rdf:typeDataset Domain[6]
Rdf:typeDataset Domain[8]
Raw Doremi Weight0.13[7]
Raw Doremi Weight0.119[8]
Sampling Weight0.018[7]
Sampling Weight0.008[8]
Had Sampling Weight0.018[1]
Exceeds Median Exposuresignificantly[1]
Had Epoch Exposure0.675[1]
Had Raw Do Re Mi Weight0.130[1]
Has Raw Do Re Mi Weight0.119[2]
Has Sampling Weight0.008[2]
Has Exposure~0.85[2]
Exposure Capped SamplingDue to 0.85[2]
Has Sampling Weight at Step17500[3]
Has Ema Loss at Step17501.335[3]
Has Raw Value at Step17500.12[3]
Has Weight Direction at Step1750↓[3]
Wrapped Times11[4]
Has Rows218[4]
Overexposed in Tiny RunTrue[4]
Exposure Wrap Count2.8[5]
Has Sampling Methoduniform[5]
Identified As Review Itemtrue[5]
Needs ConstraintTiny Bucket Cap[5]
Needs RuleMerge Rule[5]
Rank at Step7503[5]
Weight at Step7500.066[5]
Wrap Count11[6]
Exposure Frequencytoo much repeated exposure[6]
Domain Nameprofessional_benchmark[7]
Sampling Weight CauseHigh Exposure Penalty[7]
Epoch Exposure0.675[7]
Has Exposure Level0.85[8]

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.

38 undated
—
hadSamplingWeightblah/watt-activation/part-707
0.018
—
exceedsMedianExposureblah/watt-activation/part-707
significantly
—
hadEpochExposureblah/watt-activation/part-707
0.675
—
hadRawDoReMiWeightblah/watt-activation/part-707
0.130
—
hasRawDoReMiWeightblah/watt-activation/part-709
0.119
—
hasSamplingWeightblah/watt-activation/part-709
0.008
—
hasExposureblah/watt-activation/part-709
~0.85
—
exposureCappedSamplingblah/watt-activation/part-709
ex:due-to-0.85
—
hasSamplingWeightAtStep1750blah/watt-activation/part-711
0
—
hasEmaLossAtStep1750blah/watt-activation/part-711
1.335
—
hasRawValueAtStep1750blah/watt-activation/part-711
0.12
—
hasWeightDirectionAtStep1750blah/watt-activation/part-711
↓
—
wrappedTimesblah/watt-activation/part-706
11
—
hasRowsblah/watt-activation/part-706
218
—
overexposedInTinyRunblah/watt-activation/part-706
ex:true
—
rowCountblah/watt-activation/702
218
—
exposureWrapCountblah/watt-activation/702
2.8
—
hasSamplingMethodblah/watt-activation/702
uniform
—
identifiedAsReviewItemblah/watt-activation/702
true
—
needsConstraintblah/watt-activation/702
ex:tiny-bucket-cap
—
needsRuleblah/watt-activation/702
ex:merge-rule
—
rankAtStep750blah/watt-activation/702
3
—
weightAtStep750blah/watt-activation/702
0.066
—
labelblah/watt-activation/703
professional_benchmark domain
—
typeblah/watt-activation/703
ex:DatasetDomain
—
rowCountblah/watt-activation/703
218
—
wrapCountblah/watt-activation/703
11
—
exposureFrequencyblah/watt-activation/703
too much repeated exposure
—
domainNameblah/watt-activation/704
professional_benchmark
—
rawDoremiWeightblah/watt-activation/704
0.13
—
samplingWeightblah/watt-activation/704
0.018
—
samplingWeightCauseblah/watt-activation/704
ex:high-exposure-penalty
—
epochExposureblah/watt-activation/704
0.675
—
labelblah/watt-activation/706
professional_benchmark
—
typeblah/watt-activation/706
ex:DatasetDomain
—
rawDoremiWeightblah/watt-activation/706
0.119
—
samplingWeightblah/watt-activation/706
0.008
—
hasExposureLevelblah/watt-activation/706
0.85

References (8)

8 references
  1. [1]Part 7074 facts
    ctx:discord/blah/watt-activation/part-707
  2. [2]Part 7094 facts
    ctx:discord/blah/watt-activation/part-709
  3. [3]Part 7114 facts
    ctx:discord/blah/watt-activation/part-711
  4. [4]Part 7063 facts
    ctx:discord/blah/watt-activation/part-706
  5. [5]7028 facts
    ctx:discord/blah/watt-activation/702
    • full textwatt-activation-702
      text/plain3 KBdoc:agent/watt-activation-702/8ca0f2a3-b72b-46da-95b4-f4cb77d7241f
      Show excerpt
      [2026-05-01 19:32] xenonfun: **TLDR: need multithreaded and prefetching in the loader** At step 110: still stable, BPB noisy but centered roughly mid-1s so far. Token rate has crept to ~4.9K tok/s after startup. It will checkpoint at step 2…
  6. [6]7035 facts
    ctx:discord/blah/watt-activation/703
    • full textwatt-activation-703
      text/plain3 KBdoc:agent/watt-activation-703/c2aff1b4-7f23-419c-b531-687b267a9440
      Show excerpt
      [2026-05-01 20:40] xenonfun: 1K smoke finished and inference is done. Run: `/home/ms/data/clifford_ssm/checkpoints/e24_chinchilla_from_e23_best_prefetch_1k/latest.pt` Final training stats: - step `1000` - loss / BPB: `0.9332 / 1.3463` - t…
  7. [7]7045 facts
    ctx:discord/blah/watt-activation/704
    • full textwatt-activation-704
      text/plain3 KBdoc:agent/watt-activation-704/f1adf805-6f9b-4ec3-8cd5-4fef33d74eac
      Show excerpt
      [2026-05-01 21:57] xenonfun: what it had just finished and was ready: Done. I added the exposure-aware DoReMi governor with a heavy repeat penalty after one epoch/wrap. Core behavior is in [experiments/e7_zinit_depth.py](/home/ms/harmonicr…
  8. [8]7065 facts
    ctx:discord/blah/watt-activation/706
    • full textwatt-activation-706
      text/plain2 KBdoc:agent/watt-activation-706/6633be5c-fab3-495f-b5b5-8dadec7f0fbb
      Show excerpt
      [2026-05-01 23:01] xenonfun: Yes. The 5K gate run is launched and behaving well so far. Current run: `e24_chinchilla_latest_governed_5k` It is warm-started from e23 best with: - `B=20`, `T=2048` - updated Chinchilla dataset, `6,123,670` t…

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.

↑