WireLM
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
WireLM has 27 facts recorded in Dontopedia across 5 references, with 2 live disagreements.
Mostly:rdf:type(2), achieves percent better per byte information content(1), has bpb(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (7)
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.
isRunOfIs Run of(2)
- Wire Lm 40k Run
ex:wire-lm-40k-run - Wire Lm 80k Run
ex:wire-lm-80k-run
appliesToModelApplies to Model(1)
- Dyn H4 Fix
ex:dyn-h4-fix
extendsExtends(1)
- Resonant Wire Lm
ex:resonant-wire-lm
hasModelHas Model(1)
- Session 2026 03 21
ex:session-2026-03-21
hasTinyParamsRelativeToHas Tiny Params Relative to(1)
- Config 7
ex:config-7
implementsModelImplements Model(1)
- Rust Project
ex:rust-project
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 |
|---|---|---|
| Rdf:type | Language Model | [4] |
| Rdf:type | Model | [5] |
| Achieves Percent Better Per Byte Information Content | 8 | [1] |
| Has Bpb | 1.572 | [1] |
| Has Three Times Fewer Params Than | Spectral Plus Lohe V3 Bpe 8k | [1] |
| Is Byte Level Model | null | [1] |
| Uses Quaternion Architecture | null | [1] |
| Has Params Count | 8500000 | [2] |
| Serves As Size Baseline | Sweep | [2] |
| Relies on Lohe Components | attention + FFN | [3] |
| Uses Attention | Lohe Complex Split Attention | [3] |
| Uses Ffn | Lohe V3 Ffn | [3] |
| Has Parameter Count | 8250000 | [4] |
| Measured Performance in Nats Per Byte | 1.09 | [4] |
| Trained for Steps | 20000 | [4] |
| Outperforms on Metric | Spectral Lohe V3 Bpe 8k | [4] |
| Has Fewer Parameters Than | Spectral Lohe V3 Bpe 8k | [4] |
| Has Three Times Fewer Parameters Than | Spectral Lohe V3 Bpe 8k | [4] |
| Has Improved Information Content by Percentage | 8 | [4] |
| Predicts Tokens Per Byte | 1 | [4] |
| Has Performance Slope | -0.052 | [4] |
| Measured Performance in Bpb | 1.572 | [4] |
| Has Cache Type | Attention step() states | [5] |
| Has Generation Method | Token-by-token | [5] |
| Has Speed | O(1)/token | [5] |
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 (5)
ctx:discord/blah/watt-activation/part-327ctx:discord/blah/watt-activation/part-388ctx:discord/blah/watt-activation/part-425ctx:discord/blah/watt-activation/325- full textwatt-activation-325text/plain3 KB
doc:agent/watt-activation-325/2c3de55b-7643-47a7-9671-90e12c7ea3baShow excerpt
[2026-03-15 04:18] xenonfun: ``` ⏺ This is the real story: Wire LM (8.25M) is beating the best BPE model (24.47M) on per-byte CE: ┌─────────────────────────┬────────┬───────────┬───────┐ │ Model │ Params │ nats/byt…
ctx:discord/blah/watt-activation/484- full textwatt-activation-484text/plain2 KB
doc:agent/watt-activation-484/0f2365f0-484f-40bd-bbda-2a0aa92a1109Show excerpt
[2026-03-22 01:01] xenonfun: ``` ⏺ Pushed. All models now have cached inference: ┌────────────────────┬───────────────────────────┬────────────────┬───────────────────┐ │ Model │ Cache Type │ Generation …
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.