Dontopedia

Inference script

From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)

Inference script has 22 facts recorded in Dontopedia across 7 references, with 4 live disagreements.

22 facts·12 predicates·7 sources·4 in dispute

Mostly:rdf:type(4), has parameter(4), has parameter value(3)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound 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.

intendsToFixIntends to Fix(1)

intendsToWriteIntends to Write(1)

Other facts (20)

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.

20 facts
PredicateValueRef
Rdf:typeScript[3]
Rdf:typeScript File[4]
Rdf:typeScript[5]
Rdf:typeCode File[7]
Has Parameter--ckpt-dir[4]
Has Parameter--prompt[4]
Has Parameter--max-tokens[4]
Has Parameter--no-stop[4]
Has Parameter Valuecheckpoints/fineweb_e2[4]
Has Parameter ValueIs Kant kinda a cunt?[4]
Has Parameter Value300[4]
Presupposes Compiled Modelnull[1]
Receives AdditionImage Conditioned Text Generation[2]
EnablesImage Conditioned Text Generation[2]
Has Pathscripts/infer_cl100k.py[4]
Invoked Viauv run[4]
Has Featureauto-detect[6]
Requiresconfig.json[6]
Was Patchedtrue[7]
Fixed Loading Behaviortrue[7]

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.

presupposesCompiledModelblah/watt-activation/part-170
null
receivesAdditionblah/watt-activation/part-245
ex:image-conditioned-text-generation
enablesblah/watt-activation/part-245
ex:image-conditioned-text-generation
typeblah/watt-activation/58
ex:Script
labelblah/watt-activation/151
Inference script
typeblah/watt-activation/151
ex:ScriptFile
hasPathblah/watt-activation/151
scripts/infer_cl100k.py
invokedViablah/watt-activation/151
uv run
hasParameterblah/watt-activation/151
--ckpt-dir
hasParameterValueblah/watt-activation/151
checkpoints/fineweb_e2
hasParameterblah/watt-activation/151
--prompt
hasParameterValueblah/watt-activation/151
Is Kant kinda a cunt?
hasParameterblah/watt-activation/151
--max-tokens
hasParameterValueblah/watt-activation/151
300
hasParameterblah/watt-activation/151
--no-stop
typeblah/watt-activation/244
ex:Script
hasFeatureblah/watt-activation/364
auto-detect
requiresblah/watt-activation/364
config.json
labelblah/watt-activation/703
Inference scripts
typeblah/watt-activation/703
ex:CodeFile
wasPatchedblah/watt-activation/703
true
fixedLoadingBehaviorblah/watt-activation/703
true

References (7)

7 references
  1. [1]Part 1701 fact
    ctx:discord/blah/watt-activation/part-170
  2. [2]Part 2452 facts
    ctx:discord/blah/watt-activation/part-245
  3. [3]581 fact
    ctx:discord/blah/watt-activation/58
    • full textwatt-activation-58
      text/plain3 KBdoc:agent/watt-activation-58/2b893cf3-b6f9-4fdb-a94d-202a521b9459
      Show excerpt
      [2026-03-07 09:29] xenonfun: ``` Anchor v3 running. On memory — the anchor overhead is minimal: Anchor v3 extra parameters per layer (m=32): - anchor_re/im: 32 × 4 phases = 256 floats - anchor_omega: 256 floats - W_anchor_q: 320 ×
  4. [4]15111 facts
    ctx:discord/blah/watt-activation/151
    • full textwatt-activation-151
      text/plain2 KBdoc:agent/watt-activation-151/765c248d-bea7-4461-8654-f0146f5f2e83
      Show excerpt
      [2026-03-09 15:51] xenonfun: ``` Prompt: 'Is Kant kinda a cunt?' temp=0.8 top_k=40 stop=<|endoftext|> (100257) ──────────────────────────────────────────────────────────── Is Kant kinda a cunt? ───────────────────────────────────────────
  5. [5]2441 fact
    ctx:discord/blah/watt-activation/244
    • full textwatt-activation-244
      text/plain3 KBdoc:agent/watt-activation-244/12f61b26-af40-4e33-a8d7-716f2405dc1b
      Show 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
  6. [6]3642 facts
    ctx:discord/blah/watt-activation/364
    • full textwatt-activation-364
      text/plain2 KBdoc:agent/watt-activation-364/c3dc2d47-3059-4d01-bfe8-63af919a4f72
      Show excerpt
      [2026-03-17 21:30] xenonfun: ```Excellent result. This confirms that retrieval amplitude and retrieval distance-decay are separate phenomena. Interpretation: - The absolute retrieval level at seq=1024 is ~50%, likely due to undertraining /
  7. [7]7034 facts
    ctx:discord/blah/watt-activation/703
    • full textwatt-activation-703
      text/plain3 KBdoc:agent/watt-activation-703/c2aff1b4-7f23-419c-b531-687b267a9440
      Show excerpt
      [2026-05-01 20:40] xenonfun: 1K smoke finished and inference is done. Run: `/home/ms/data/clifford_ssm/checkpoints/e24_chinchilla_from_e23_best_prefetch_1k/latest.pt` Final training stats: - step `1000` - loss / BPB: `0.9332 / 1.3463` - t

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.