Kan Iter 10
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Kan Iter 10 has 21 facts recorded in Dontopedia across 2 references.
Mostly:has learning rate(2), has loss(2), has tokens per second(2)
Maturity scale
raw canonical shape-checked rule-derived certifiedOther facts (21)
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 Learning Rate | 0.0003 | [1] |
| Has Learning Rate | 0.0003 | [2] |
| Has Loss | 5.4382 | [1] |
| Has Loss | 5.4382 | [2] |
| Has Tokens Per Second | 14738 | [1] |
| Has Tokens Per Second | 14738 | [2] |
| Temporally Precedes | Kan Iter 20 | [1] |
| Has Gpu Cache Gb | 14 | [1] |
| Has Gpu Peak Gb | 14 | [1] |
| Has Higher Loss Than | Kan Iter 20 | [1] |
| Has Iteration Number | 10 | [1] |
| Has Process Rssgb | 3.2 | [1] |
| Is Part of | Kan Run | [1] |
| Has Gpu Active Gb | 1 | [1] |
| Rdf:type | Iteration | [2] |
| Iteration Number | 10 | [2] |
| Part of | Kan Run | [2] |
| Has Gpu Memory Active | 1.0GB | [2] |
| Has Gpu Memory Peak | 14.0GB | [2] |
| Has Gpu Memory Cache | 14.0GB | [2] |
| Has Process Memory | 3.2GB | [2] |
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)
ctx:discord/blah/watt-activation/part-112ctx:discord/blah/watt-activation/112- full textwatt-activation-112text/plain3 KB
doc:agent/watt-activation-112/dad4c21d-6073-4542-b09b-5b5162617d30Show excerpt
[2026-03-08 22:40] xenonfun: ``` === kan (seq=2048) === Training HarmonicGPT | attn=kan d=768 L=12 H=12 seq=2048 Parameters: 86,826,048 Loaded 5,000 sequences of length 2048 iter 10/100 | loss 5.4382 | lr 3.00e-04 | tok/s 14,738 | gpu 1…
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.