Ppl Metric
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Ppl Metric has 10 facts recorded in Dontopedia across 8 references, with 1 live disagreement.
Mostly:measures model quality(2), rdf:type(2), is primary evaluation(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (4)
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.
hasMetricHas Metric(1)
- Training Complete Event
ex:training-complete-event
isHealthyIs Healthy(1)
- Code Lm
ex:code-lm
presupposesLowerPplIsBetterPresupposes Lower Ppl Is Better(1)
- Anchor Method
ex:anchor-method
referencesPerplexityLossMetricReferences Perplexity Loss Metric(1)
- Ppl
ex:ppl
Other facts (10)
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 |
|---|---|---|
| Measures Model Quality | Language Model | [4] |
| Measures Model Quality | Training Session | [5] |
| Rdf:type | Metric | [7] |
| Rdf:type | Metric | [8] |
| Is Primary Evaluation | Sweep | [1] |
| Measures Performance | Configs | [2] |
| Decreases Over Iters | Iters 30500 to 37000 | [3] |
| Decreases Over Versions | 80.3 to 75.1 | [6] |
| Has Value | 83.4 | [8] |
| Is Perplexity | Language Model | [8] |
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 (8)
ctx:discord/blah/watt-activation/part-45ctx:discord/blah/watt-activation/part-48ctx:discord/blah/watt-activation/part-97ctx:discord/blah/watt-activation/part-127ctx:discord/blah/watt-activation/part-161ctx:discord/blah/watt-activation/part-251ctx:discord/blah/watt-activation/229ctx:discord/blah/watt-activation/244- full textwatt-activation-244text/plain3 KB
doc:agent/watt-activation-244/12f61b26-af40-4e33-a8d7-716f2405dc1bShow excerpt
[2026-03-12 05:23] xenonfun: ❯ can we infer on images and audio or get them back out? ⏺ Not yet — the current architecture is encoder-only for image/audio (projects them into the sequence for cross-modal context), but only has a text outpu…
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.