Step 100

as donto believed on

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

Step 100 has 86 facts recorded in Dontopedia across 15 references, with 8 live disagreements.

86 facts·52 predicates·15 sources·8 in dispute

Mostly:has loss(11), has ppl(8), precedes(5)

Has Lossin dispute`hasLoss`

11.9294^[1]all time · Part 125
19.6886^[3]all time · Part 189
11.6429^[4]all time · Part 210
11.9301^[5]all time · Part 202
11.8396^[6]all time · Part 211
11.6335^[7]all time · Part 203
11.8^[8]all time · Part 228
9.2353^[9]all time · Part 216
11.8134^[10]all time · Part 230
4.9062^[11]all time · Part 331

Inbound mentions (9)

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.

`associatedWithStep`Associated With Step(1)

Eval Checkpoint 1ex:eval-checkpoint-1

`atStep`At Step(1)

Lohe Optimizerex:lohe-optimizer

`hasStep`Has Step(1)

Experiment Lohe Spherical Lohe V2 Run2ex:experiment-lohe-spherical-lohe-v2-run2

`improvedFrom`Improved From(1)

Val Lossex:val-loss

`lossDecreasesFromStep100To200`Loss Decreases From Step100 To200(1)

Training Runex:training-run

`occursBetween`Occurs Between(1)

Gelation Onsetex:gelation-onset

`occursBetweenSteps`Occurs Between Steps(1)

Gelation Transitionex:gelation-transition

`succeeds`Succeeds(1)

Step 150ex:step-150

`usesArrowLabels`Uses Arrow Labels(1)

Xenonfunex:xenonfun

Other facts (74)

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.

74 facts

Predicate	Value	Ref
Has Ppl	151666.9	[1]
Has Ppl	350.1	[2]
Has Ppl	355350718.4	[3]
Has Ppl	113884.2	[4]
Has Ppl	151765.5	[5]
Has Ppl	138637.2	[6]
Has Ppl	10253.1	[9]
Has Ppl	135046.4	[10]
Precedes	Step 200	[2]
Precedes	Step 200	[4]
Precedes	Step 200	[5]
Precedes	Step 200	[8]
Precedes	Step 200	[11]
Has Tok Per S	17254	[3]
Has Tok Per S	13027	[4]
Has Tok Per S	7953	[5]
Has Tok Per S	14804	[10]
Has Elapsed Time	50s	[5]
Has Elapsed Time	52s	[6]
Has Elapsed Time	68s	[9]
Has Elapsed Time	30	[15]
Has Tokens Per Second	12084	[6]
Has Tokens Per Second	104253	[11]
Has Tokens Per Second	14796	[15]
Has Attn R	0.5845	[5]
Has Attn R	0.5846	[7]
Has Tok Per Sec	14853	[7]
Has Tok Per Sec	6324	[9]
Has Perplexity	135.1	[11]
Has Perplexity	133039.8	[15]
Step Number	100	[13]
Step Number	100	[15]
Tok Per S	12894	[1]
Has Lr	2.00e-05	[1]
Progress	0.6%	[1]
Step Time	635ms	[1]
Eta	175min	[1]
Has Val Loss	5.8583	[2]
Elapsed	25	[3]
Has Sys R	0.345	[4]
Has Kappa	0.177	[4]
Has Time	32s	[4]
Has Overhead	+129ms	[5]
Has Step Progress	100/1000	[5]
Has Extra Time	+39ms	[7]
Has Ppl	112817.7	[7]
Has Elapsed Secs	28	[7]
Has B9 R	0.363	[8]
Has B10 Minus B11	-0.014	[8]
Has B10 R	0.347	[8]
Has B11 R	0.362	[8]
Has B9 Disorder	-0.016	[8]
Has Flat Output Blocks	Output Blocks	[8]
Is Disordered	System	[8]
Is Symmetric	System	[8]
System Is Liquid	true	[8]
System Is Symmetric	true	[8]
Succeeds	Step 50	[10]
Has Elapsed	31s	[10]
Part of Training	Training Run	[10]
Has Ms Per Step	157	[11]
Has Learning Rate	6.00e-05	[11]
Is at Progress	0.5%	[11]
Has Eta	52min	[11]
Has Rollout Steps	0.04	[11]
Observes Governor Effect	Professional Benchmark	[12]
Rdf:type	Training Step	[14]
Loss Value	11.6335	[14]
Perplexity Value	112817.7	[14]
Tokens Per Second	14853	[14]
Elapsed Time	28s	[14]
Attn R Value	0.5846	[14]
Step Overhead	+39ms	[14]
Part of	Training Run 20260311 001319	[15]

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.

86 undated

—

tokPerSblah/watt-activation/part-125

12894

—

hasPplblah/watt-activation/part-125

151666.9

—

hasLrblah/watt-activation/part-125

2.00e-05

—

hasLossblah/watt-activation/part-125

11.9294

—

progressblah/watt-activation/part-125

0.6%

—

stepTimeblah/watt-activation/part-125

635ms

—

etablah/watt-activation/part-125

175min

—

hasPplblah/watt-activation/part-196

350.1

—

precedesblah/watt-activation/part-196

ex:step-200

—

hasValLossblah/watt-activation/part-196

5.8583

—

hasPplblah/watt-activation/part-189

355350718.4

—

hasLossblah/watt-activation/part-189

19.6886

—

hasTokPerSblah/watt-activation/part-189

17254

—

elapsedblah/watt-activation/part-189

—

hasTokPerSblah/watt-activation/part-210

13027

—

hasLossblah/watt-activation/part-210

11.6429

—

precedesblah/watt-activation/part-210

ex:step-200

—

hasSysRblah/watt-activation/part-210

0.345

—

hasPplblah/watt-activation/part-210

113884.2

—

hasKappablah/watt-activation/part-210

0.177

—

hasTimeblah/watt-activation/part-210

32s

—

hasAttnRblah/watt-activation/part-202

0.5845

—

hasElapsedTimeblah/watt-activation/part-202

50s

—

precedesblah/watt-activation/part-202

ex:step-200

—

hasLossblah/watt-activation/part-202

11.9301

—

hasTokPerSblah/watt-activation/part-202

7953

—

hasPplblah/watt-activation/part-202

151765.5

—

hasOverheadblah/watt-activation/part-202

+129ms

—

hasStepProgressblah/watt-activation/part-202

100/1000

—

hasElapsedTimeblah/watt-activation/part-211

52s

—

hasLossblah/watt-activation/part-211

11.8396

—

hasPplblah/watt-activation/part-211

138637.2

—

hasTokensPerSecondblah/watt-activation/part-211

12084

—

hasLossblah/watt-activation/part-203

11.6335

—

hasTokPerSecblah/watt-activation/part-203

14853

—

hasExtraTimeblah/watt-activation/part-203

+39ms

—

hasPPLblah/watt-activation/part-203

112817.7

—

hasElapsedSecsblah/watt-activation/part-203

—

hasAttnRblah/watt-activation/part-203

0.5846

—

hasB9Rblah/watt-activation/part-228

0.363

—

hasB10MinusB11blah/watt-activation/part-228

-0.014

—

hasB10Rblah/watt-activation/part-228

0.347

—

hasB11Rblah/watt-activation/part-228

0.362

—

hasB9Disorderblah/watt-activation/part-228

-0.016

—

hasFlatOutputBlocksblah/watt-activation/part-228

ex:output-blocks

—

hasLossblah/watt-activation/part-228

11.8

—

isDisorderedblah/watt-activation/part-228

ex:system

—

isSymmetricblah/watt-activation/part-228

ex:system

—

precedesblah/watt-activation/part-228

ex:step-200

—

systemIsLiquidblah/watt-activation/part-228

true

—

systemIsSymmetricblah/watt-activation/part-228

true

—

hasElapsedTimeblah/watt-activation/part-216

68s

—

hasLossblah/watt-activation/part-216

9.2353

—

hasTokPerSecblah/watt-activation/part-216

6324

—

hasPplblah/watt-activation/part-216

10253.1

—

hasLossblah/watt-activation/part-230

11.8134

—

hasTokPerSblah/watt-activation/part-230

14804

—

hasPplblah/watt-activation/part-230

135046.4

—

succeedsblah/watt-activation/part-230

ex:step-50

—

hasElapsedblah/watt-activation/part-230

31s

—

partOfTrainingblah/watt-activation/part-230

ex:training-run

—

hasMsPerStepblah/watt-activation/part-331

157

—

hasTokensPerSecondblah/watt-activation/part-331

104253

—

hasLossblah/watt-activation/part-331

4.9062

—

hasLearningRateblah/watt-activation/part-331

6.00e-05

—

isAtProgressblah/watt-activation/part-331

0.5%

—

hasEtablah/watt-activation/part-331

52min

—

precedesblah/watt-activation/part-331

ex:step-200

—

hasRolloutStepsblah/watt-activation/part-331

0.04

—

hasPerplexityblah/watt-activation/part-331

135.1

—

observesGovernorEffectblah/watt-activation/part-707

ex:professional-benchmark

—

stepNumberblah/watt-activation/13

100

—

typeblah/watt-activation/202

ex:TrainingStep

—

lossValueblah/watt-activation/202

11.6335

—

perplexityValueblah/watt-activation/202

112817.7

—

tokensPerSecondblah/watt-activation/202

14853

—

elapsedTimeblah/watt-activation/202

28s

—

attnRValueblah/watt-activation/202

0.5846

—

stepOverheadblah/watt-activation/202

+39ms

—

labelblah/watt-activation/214

Step 100

—

stepNumberblah/watt-activation/214

100

—

partOfblah/watt-activation/214

ex:training-run-20260311-001319

—

hasLossblah/watt-activation/214

11.7984

—

hasPerplexityblah/watt-activation/214

133039.8

—

hasTokensPerSecondblah/watt-activation/214

14796

—

hasElapsedTimeblah/watt-activation/214

References (15)

15 references

[1]Part 1257 facts
ctx:discord/blah/watt-activation/part-125
[2]Part 1963 facts
ctx:discord/blah/watt-activation/part-196
[3]Part 1894 facts
ctx:discord/blah/watt-activation/part-189
[4]Part 2107 facts
ctx:discord/blah/watt-activation/part-210
[5]Part 2028 facts
ctx:discord/blah/watt-activation/part-202
[6]Part 2114 facts
ctx:discord/blah/watt-activation/part-211
[7]Part 2036 facts
ctx:discord/blah/watt-activation/part-203
[8]Part 22812 facts
ctx:discord/blah/watt-activation/part-228
[9]Part 2164 facts
ctx:discord/blah/watt-activation/part-216
[10]Part 2306 facts
ctx:discord/blah/watt-activation/part-230
[11]Part 3319 facts
ctx:discord/blah/watt-activation/part-331
[12]Part 7071 fact
ctx:discord/blah/watt-activation/part-707
[13]131 fact
ctx:discord/blah/watt-activation/13
- full textwatt-activation-13
  text/plain2 KBdoc:agent/watt-activation-13/8fb2ebd5-d74a-4511-9dac-1a7047d803b2
  Show excerpt
  [2026-02-28 18:28] xenonfun: 2026-02-28 13:23:39,591 [INFO] cross_species_1b.finetune: Step 10/1000: train_loss=7.9112, lr=2.20e-06, 1051 tok/s **1000 steps ≈ 8.5 hours.** [2026-02-28 18:31] xenonfun: (files: Screenshot_2026-02-28_at_…
[14]2027 facts
ctx:discord/blah/watt-activation/202
- full textwatt-activation-202
  text/plain3 KBdoc:agent/watt-activation-202/bfa1acbb-7492-43ef-befc-99b62acad42a
  Show excerpt
  [2026-03-10 06:54] xenonfun: ``` lohe_spherical+lohe_v2 summary: train loss (last 50): 7.8510 ppl=2568.3 val loss (final): 7.8019 ppl=2445.1 sys_r @1000: 0.8817 global_r @1000: 0.2008 attn_r @1000:…
[15]2147 facts
ctx:discord/blah/watt-activation/214
- full textwatt-activation-214
  text/plain2 KBdoc:agent/watt-activation-214/6ebedb80-7ad3-4d5e-8432-4c87da7e9ad5
  Show excerpt
  [2026-03-11 04:17] xenonfun: ``` Model: lohe_spherical+lohe_v3 768d 12L 12H vocab=100277 Steps: 1000 BS=4 SEQ=1024 (4.1M tokens/run) Phase metrics sidecar → logs/phase_metrics_lohe_spherical_20260311_001319.jsonl ====================…

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.

Step 100

Has Lossin disputehasLoss

Inbound mentions (9)

associatedWithStepAssociated With Step(1)

atStepAt Step(1)

hasStepHas Step(1)

improvedFromImproved From(1)

lossDecreasesFromStep100To200Loss Decreases From Step100 To200(1)

occursBetweenOccurs Between(1)

occursBetweenStepsOccurs Between Steps(1)

succeedsSucceeds(1)

usesArrowLabelsUses Arrow Labels(1)

Other facts (74)

Timeline

References (15)

See also

Keep researching

Has Lossin dispute`hasLoss`

`associatedWithStep`Associated With Step(1)

`atStep`At Step(1)

`hasStep`Has Step(1)

`improvedFrom`Improved From(1)

`lossDecreasesFromStep100To200`Loss Decreases From Step100 To200(1)

`occursBetween`Occurs Between(1)

`occursBetweenSteps`Occurs Between Steps(1)

`succeeds`Succeeds(1)

`usesArrowLabels`Uses Arrow Labels(1)