Training Step 500
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Training Step 500 has 25 facts recorded in Dontopedia across 6 references, with 1 live disagreement.
Mostly:has loss(3), rdf:type(2), is out of(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (2)
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.
decreasesOverStepsDecreases Over Steps(1)
- Training Loss
ex:training-loss
precedesPrecedes(1)
- Params Print
ex:params-print
Other facts (25)
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 Loss | 2.6037 | [1] |
| Has Loss | 2.5678 | [2] |
| Has Loss | 6.1738 | [3] |
| Rdf:type | Training Step | [5] |
| Rdf:type | Training Step | [6] |
| Is Out of | 2000 | [1] |
| Precedes | Training Step 1000 | [2] |
| Has Step Number | 500 | [2] |
| Has Ppl | 480 | [3] |
| Has Progress Percent | 3.0% | [3] |
| Is Step Number | 500 | [3] |
| Has Duration Ms | 666 | [3] |
| Has Eta Minutes | 180 | [3] |
| Has Lr | 0.00005 | [3] |
| Has Tokens Per Second | 12296 | [3] |
| Common Endpoint | Sweeps | [4] |
| Step Number | 500 | [5] |
| Reported Loss | 2.5678 | [5] |
| Step Index | 500 | [6] |
| Loss Value | 8.2888 | [6] |
| Perplexity | 3979.1 | [6] |
| Tokens Per Second | 8580 | [6] |
| Elapsed Time | 252s | [6] |
| Attention Ratio | 0.6695 | [6] |
| Has Overhead Duration | 127ms | [6] |
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 (6)
ctx:discord/blah/vidya/part-4ctx:discord/blah/vidya/part-3ctx:discord/blah/watt-activation/part-147ctx:discord/blah/watt-activation/part-392ctx:discord/blah/vidya/3- full textvidya-3text/plain2 KB
doc:agent/vidya-3/36dc77aa-7dff-487d-af09-d932dc181732Show excerpt
[2026-02-20 07:37] rolandnsharp7643: roland@cube:~/code/flow/ocaml$ ./microgpt_scaled num docs: 16478 vocab size: 80 num params: 206848 step 500 / 2000 | loss 2.5678 step 1000 / 2000 | loss 2.1392 step 1500 / 2000 | loss 1.7365 step 2000 /…
ctx:discord/blah/watt-activation/201- full textwatt-activation-201text/plain3 KB
doc:agent/watt-activation-201/d56596e8-50bd-4e0f-82d6-460f54290418Show excerpt
[2026-03-10 06:49] xenonfun: ``` ====================================================================== CONFIG: lohe_spherical+lohe_v2 ====================================================================== Parameters: 93,766,464 step …
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.