e23
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
e23 has 20 facts recorded in Dontopedia across 2 references.
Mostly:believed superior to prior runs(1), expected to hit bpb by step1000(1), expected to hit bpb less than1point8 by step1500(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.
statesModelDoesExpectedWorkPerStepStates Model Does Expected Work Per Step(1)
- Post 2026 05 01 03 44
ex:post-2026-05-01-03-44
Other facts (19)
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 |
|---|---|---|
| Believed Superior to Prior Runs | E22 E22c E18 | [1] |
| Expected to Hit Bpb by Step1000 | ~2.0 | [1] |
| Expected to Hit Bpb Less Than1point8 by Step1500 | <1.8 | [1] |
| Has30 Percent More Active Parameters Per Token Than | E22 Model | [1] |
| Has40 Percent More Total Parameters Than | E22 Model | [1] |
| Has Active Parameters Per Token | 40M | [1] |
| Has Freed8bit Adam Memory Headroom | Training Optimizer | [1] |
| Has Projected Bpb at Step1k | ~2.1 | [1] |
| Has Projected Bpb at Step25k | ~1.30-1.35 | [1] |
| Has Projected Bpb at Step5k | ~1.7 | [1] |
| Has Projected Final Val Bpb If Slope Flattens50percent | ~1.35-1.40 | [1] |
| Has T4k Capable Architecture | when data arrives | [1] |
| Has Total Parameters | 125M | [1] |
| Has Wall Time | 5.2h | [1] |
| Has Wall Time1point4x That of | E22 Model | [1] |
| Improves Over E22 | In Parameters and Active Params | [1] |
| Is Ahead of All Prior Runs at Matched Step If Condition | <1.8 by step 1.5K | [1] |
| On Track If Current Slope Holds | Current Training Slope | [1] |
| Rdf:type | Model Architecture | [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-700ctx:discord/blah/watt-activation/697- full textwatt-activation-697text/plain3 KB
doc:agent/watt-activation-697/0bcf6f97-e39d-494c-8129-e73c0ac78f73Show excerpt
[2026-05-01 03:28] xenonfun: ``` step | bpb | loss | tok/s | iter/s | elapsed -------+--------+---------+------------+---------+---------- 1 | 5.809 | 4.0263 | 576 | 0.04 | 0m28s 50 | 4.926 | …
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.