Dontopedia

Experiment 3: Multi-Seed MNIST (5 seeds, 1k/500)

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

Experiment 3: Multi-Seed MNIST (5 seeds, 1k/500) has 15 facts recorded in Dontopedia across 5 references, with 1 live disagreement.

15 facts·12 predicates·5 sources·1 in dispute

Mostly:has seed count(2), has result metric(2), is run hybrid byte code training(1)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (6)

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.

measuredInMeasured in(4)

handlesExperimentHandles Experiment(1)

isRunHybridByteCodeTrainingIs Run Hybrid Byte Code Training(1)

Other facts (14)

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.

14 facts
PredicateValueRef
Has Seed Count5[4]
Has Seed Count5[5]
Has Result MetricBest Accuracy[5]
Has Result MetricGelation Rate[5]
Is Run Hybrid Byte Code TrainingExperiment 3[1]
Used Venom Amount1/16 gr[2]
Killed Guinea Pig in45 minutes[2]
Has TitleExp 3: Long context behavior — seq 256/512/1024[3]
Has Sequence Lengths256/512/1024[3]
Is Type ofMulti Seed Mnist[4]
InvolvesMnist Dataset 1k 500[4]
Has Method Count3[4]
Rdf:typeExperiment[5]
Uses DatasetMnist[5]

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.

isRunHybridByteCodeTrainingblah/watt-activation/part-288
ex:experiment-3
usedVenomAmounttrove-cooktown/chinese-fishermen
1/16 gr
killedGuineaPigIntrove-cooktown/chinese-fishermen
45 minutes
hasTitleblah/watt-activation/449
Exp 3: Long context behavior — seq 256/512/1024
hasSequenceLengthsblah/watt-activation/449
256/512/1024
isTypeOfblah/watt-activation/464
ex:multi-seed-mnist
involvesblah/watt-activation/464
ex:mnist-dataset-1k-500
hasSeedCountblah/watt-activation/464
5
hasMethodCountblah/watt-activation/464
3
typeblah/watt-activation/466
ex:Experiment
labelblah/watt-activation/466
Experiment 3: Multi-Seed MNIST (5 seeds, 1k/500)
usesDatasetblah/watt-activation/466
ex:mnist
hasSeedCountblah/watt-activation/466
5
hasResultMetricblah/watt-activation/466
ex:best-accuracy
hasResultMetricblah/watt-activation/466
ex:gelation-rate

References (5)

5 references
  1. [1]Part 2881 fact
    ctx:discord/blah/watt-activation/part-288
  2. ctx:genes/trove-cooktown/chinese-fishermen
  3. [3]4492 facts
    ctx:discord/blah/watt-activation/449
    • full textwatt-activation-449
      text/plain2 KBdoc:agent/watt-activation-449/f6da4410-b0fa-46d2-9cb5-99d205dd2e55
      Show excerpt
      [2026-03-21 03:48] xenonfun: ⏺ You're right. At step 950: ``` ┌──────────┬───────────┬──────────┐ │ Metric │ Helmholtz │ v3 │ ├──────────┼───────────┼──────────┤ │ BPB │ 2.257 │ 2.257 │ ├──────────┼───────────┼
  4. [4]4644 facts
    ctx:discord/blah/watt-activation/464
    • full textwatt-activation-464
      text/plain3 KBdoc:agent/watt-activation-464/599938d0-3182-4b42-bb04-4488236f82bc
      Show excerpt
      [2026-03-21 18:08] xenonfun: ``` Key observations: - Rust achieves significantly higher accuracy (99.1% vs 89.5% best) — the GPU-accelerated training does more effective optimization per epoch - Gelation detected at the same step (12)
  5. [5]4666 facts
    ctx:discord/blah/watt-activation/466
    • full textwatt-activation-466
      text/plain3 KBdoc:agent/watt-activation-466/a3a63aa0-7a24-45e3-ab81-70bd522fcd80
      Show excerpt
      [2026-03-21 18:43] xenonfun: ``` Rust Replication Results (vs Python) Experiment 4: XOR Parameter Efficiency ┌────────────────┬─────────────────┬──────────────────┐ │ │ Python │ Rust │ ├──────

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.