Sgd Momentum
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Sgd Momentum has 24 facts recorded in Dontopedia across 2 references, with 3 live disagreements.
Mostly:has higher capacity than(2), has better snr than(2), preserves signal quality(2)
Maturity scale
raw canonical shape-checked rule-derived certifiedHas Higher Capacity Thanin disputehasHigherCapacityThan
- Rotational Adam W[1]all time · Part 459
- Rotational Adamw[2]sourceall time · 457
Has Better Snr Thanin disputehasBetterSnrThan
- Rotational Adam W[1]all time · Part 459
- Rotational Adamw[2]sourceall time · 457
Preserves Signal Qualityin disputepreservesSignalQuality
Achieves at Full ScaleachievesAtFullScale
Fails to Learn Representations BecausefailsToLearnRepresentationsBecause
- Wrong Lr for Most Parameters[1]all time · Part 459
Has CapacityhasCapacity
- 3.6[1]all time · Part 459
Has SnrhasSnr
- -4.3[1]all time · Part 459
Has Step1000 BpbhasStep1000Bpb
- 3.833[1]all time · Part 459
Uses Uniform LrusesUniformLr
- null[1]all time · Part 459
Worse Bpb ThanworseBpbThan
- Rotational Adam W[1]all time · Part 459
Lower Bpb Indicates BetterlowerBpbIndicatesBetter
- null[1]all time · Part 459
Has Wrong Learning Rate for Most ParametershasWrongLearningRateForMostParameters
- true[2]sourceall time · 457
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.
comparesOptimizersCompares Optimizers(1)
- Step 1000 Comparison Table
ex:step-1000-comparison-table
winsOverWins Over(1)
- Rotational Adam W
ex:rotational-adam-w
winsOverByMarginWins Over by Margin(1)
- Rotational Adamw
ex:rotational-adamw
Other facts (9)
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 (2)
- custom
ctx:discord/blah/watt-activation/part-459 - custom
ctx:discord/blah/watt-activation/457- full textwatt-activation-457text/plain2 KB
doc:agent/watt-activation-457/2e8b03b4-3b3a-4823-b2c8-a92329812095Show excerpt
[2026-03-21 13:42] xenonfun: ``` ⏺ Clear answer at step 1000: ┌─────────────────┬───────────────┬─────────┬──────────┐ │ Optimizer │ Step 1000 BPB │ SNR │ Capacity │ ├─────────────────┼───────────────┼─────────┼──────────┤ …
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.