Dontopedia

Training System

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

Training System has 23 facts recorded in Dontopedia across 4 references, with 2 live disagreements.

23 facts·21 predicates·4 sources·2 in dispute

Mostly:rdf:type(2), has feature(2), achieves speedup(1)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (3)

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.

announcesFeatureAnnounces Feature(1)

describedFeatureDescribed Feature(1)

mentionsTrainingSystemMentions Training System(1)

Other facts (23)

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.

23 facts
PredicateValueRef
Rdf:typeSoftware System[3]
Rdf:typeComputational System[4]
Has FeatureParallel Tokenizer[3]
Has FeatureAsync Checkpoint Saving[3]
Achieves Speedup68[1]
Has Async Checkpoint Savingtrue[1]
Has Parallel Tokenizer With Cachingtrue[1]
Includes Philosophy Corpus Examplestrue[1]
Is Completetrue[1]
Philosophy Corpus DependentPhilosophy Corpus[1]
Supports Dora Fine Tuningtrue[1]
Supports From Scratch Training With Best Checkpointstrue[1]
Uses Learning Rate Schedulingwarmup + cosine/linear[1]
Uses Mixed Precision Fp16true[1]
Uses Optimized Mlx Conv1dtrue[1]
Involves RegionFortage Region[2]
Has OptimizationMlx Conv1d Optimization[3]
Supports ModeFrom Scratch Training[3]
Uses Learning Rate ScheduleWarmup Plus Schedule[3]
Uses PrecisionFp16[3]
Uses Fine Tuning MethodDora[3]
Driven byPacked Batch[4]
Has StateStable Regime[4]

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.

achievesSpeedupblah/watt-activation/part-18
68
hasAsyncCheckpointSavingblah/watt-activation/part-18
true
hasParallelTokenizerWithCachingblah/watt-activation/part-18
true
includesPhilosophyCorpusExamplesblah/watt-activation/part-18
true
isCompleteblah/watt-activation/part-18
true
philosophyCorpusDependentblah/watt-activation/part-18
ex:philosophy-corpus
supportsDoraFineTuningblah/watt-activation/part-18
true
supportsFromScratchTrainingWithBestCheckpointsblah/watt-activation/part-18
true
usesLearningRateSchedulingblah/watt-activation/part-18
warmup + cosine/linear
usesMixedPrecisionFp16blah/watt-activation/part-18
true
usesOptimizedMlxConv1dblah/watt-activation/part-18
true
involvesRegionblah/watt-activation/part-339
ex:fortage-region
typeblah/watt-activation/18
ex:SoftwareSystem
hasOptimizationblah/watt-activation/18
ex:mlx-conv1d-optimization
hasFeatureblah/watt-activation/18
ex:parallel-tokenizer
supportsModeblah/watt-activation/18
ex:from-scratch-training
hasFeatureblah/watt-activation/18
ex:async-checkpoint-saving
usesLearningRateScheduleblah/watt-activation/18
ex:warmup-plus-schedule
usesPrecisionblah/watt-activation/18
ex:fp16
usesFineTuningMethodblah/watt-activation/18
ex:dora
typeblah/watt-activation/249
ex:ComputationalSystem
drivenByblah/watt-activation/249
ex:packed-batch
hasStateblah/watt-activation/249
ex:stable-regime

References (4)

4 references
  1. [1]Part 1811 facts
    ctx:discord/blah/watt-activation/part-18
  2. [2]Part 3391 fact
    ctx:discord/blah/watt-activation/part-339
  3. [3]188 facts
    ctx:discord/blah/watt-activation/18
    • full textwatt-activation-18
      text/plain3 KBdoc:agent/watt-activation-18/812b86dc-f424-4d37-b858-94cd90ca9fbf
      Show excerpt
      [2026-03-04 05:47] xenonfun: feat: Complete training system with optimizations - 68x speedup with optimized MLX conv1d - Parallel tokenizer with caching - From-scratch training with best checkpoints - Async checkpoint saving - Learning rate
  4. [4]2493 facts
    ctx:discord/blah/watt-activation/249
    • full textwatt-activation-249
      text/plain2 KBdoc:agent/watt-activation-249/8436e2be-67b0-44b5-9c54-97c19c68045d
      Show excerpt
      [2026-03-12 08:24] xenonfun: ⏺ Early packed phase data — only ~550 steps in but already showing interesting dynamics: r heatmap (top-left): Much more visible step-to-step fluctuation than the unpacked runs. The vertical striping pattern

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.