Resonance Seq 2048 Run
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-05-02.)
Resonance Seq 2048 Run has 30 facts recorded in Dontopedia across 1 reference.
Mostly:achieves completion despite nan(1), existentially commits to gpu usage(1), has average tokens per second(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedAchieves Completion Despite NanachievesCompletionDespiteNan
- null[1]all time · Part 111
Existentially Commits to Gpu UsageexistentiallyCommitsToGpuUsage
- null[1]all time · Part 111
Has Average Tokens Per SecondhasAverageTokensPerSecond
- 11808[1]all time · Part 111
Has Constant Loss ValuehasConstantLossValue
- nan[1]all time · Part 111
Has Elapsed Time SecondshasElapsedTimeSeconds
- 69.4[1]all time · Part 111
Has Final Gpu Active MemoryhasFinalGpuActiveMemory
- 0.86 GB[1]all time · Part 111
Has Final Gpu Cache MemoryhasFinalGpuCacheMemory
- 17.30 GB[1]all time · Part 111
Has Final Gpu Peak MemoryhasFinalGpuPeakMemory
- 16.77 GB[1]all time · Part 111
Has Final LosshasFinalLoss
- nan[1]all time · Part 111
Has Final Process Rss MemoryhasFinalProcessRssMemory
- 3.19 GB[1]all time · Part 111
Has Gpu Cache GbhasGpuCacheGb
- 17.3[1]all time · Part 111
Has Gpu Memory Usage GbhasGpuMemoryUsageGb
- 0.9[1]all time · Part 111
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.
belongsToRunBelongs to Run(2)
- Iteration 10 Resonance
ex:iteration-10-resonance - Iteration 20 Resonance
ex:iteration-20-resonance
causesTrainingFailureCauses Training Failure(1)
- Nan Loss
ex:nan-loss
guessesNeedsFixingGuesses Needs Fixing(1)
- Xenonfun
ex:xenonfun
includesSubsectionIncludes Subsection(1)
- Seq Len 2048 Benchmark
ex:seq-len-2048-benchmark
Other facts (18)
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.
| Predicate | Value | Ref |
|---|---|---|
| Has Gpu Peak Gb | 16.8 | [1] |
| Has Head Count | 12 | [1] |
| Has Layer Count | 12 | [1] |
| Has Model Dimension | 768 | [1] |
| Has Near Constant Gpu Peak | null | [1] |
| Has Parameter Count | 73543104 | [1] |
| Has Process Gb | 3.2 | [1] |
| Has Sequence Length | 2048 | [1] |
| Has Status | complete | [1] |
| Has Total Iterations | 100 | [1] |
| Has Total Tokens Processed | 819200 | [1] |
| Loaded Sequence Count | 5000 | [1] |
| Reported With Certainty | null | [1] |
| Trains Model | Harmonic Gpt | [1] |
| Exhibits Decreasing Learning Rate | Learning Rate Schedule | [1] |
| Exhibits Increasing Tokens Per Second | null | [1] |
| Presupposes Problematic Nan Loss | null | [1] |
| Contrasts With Successful Training | null | [1] |
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 (1)
- custom
ctx:discord/blah/watt-activation/part-111
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.