Takeaway: Best current option is H=96
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Takeaway: Best current option is H=96 has 7 facts recorded in Dontopedia across 2 references.
Mostly:rdf:type(1), described by(1), attributes crash to(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (1)
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.
containsObservationContains Observation(1)
- Message 1
ex:message-1
Other facts (6)
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 |
|---|---|---|
| Rdf:type | Conclusion | [1] |
| Described by | m64 crashed — sin coupling with 64 anchors diverges | [1] |
| Attributes Crash to | sin coupling with 64 anchors diverges | [1] |
| Identifies Best Configuration | Model Config H96 | [2] |
| Justification | smaller and faster | [2] |
| Citing Reason | Alignment Reason | [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/51- full textwatt-activation-51text/plain3 KB
doc:agent/watt-activation-51/53cd7634-6f49-4bb0-9083-629d07a48d10Show excerpt
[2026-03-07 07:59] xenonfun: ``` Full results are in. Here's the complete picture: ┌───────────────────┬───────────┬──────────┬──────┬──────────┬──────┐ │ Config │ Final PPL │ Best PPL │ r │ Clusters │ it/s │ ├─────────…
ctx:discord/blah/watt-activation/677- full textwatt-activation-677text/plain2 KB
doc:agent/watt-activation-677/0a3c1f90-3dc6-48cb-b08e-24d6d4275edeShow excerpt
[2026-04-22 13:11] xenonfun: • For the H=40 preset (D=560, ~19.0M params), current measured GPU training is: ``` 21,642 tok/s at B=16, T=256 on M3 Ultra. I also tried nearby H values in multiples of 4: | H | D | Params | GPU tok/s |…
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.