Dontopedia

Adjust the Collection Dimension

From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-07.)

Adjust the Collection Dimension has 19 facts recorded in Dontopedia across 6 references, with 1 live disagreement.

19 facts·16 predicates·6 sources·1 in dispute

Mostly:rdf:type(3), relies on factual recall scaling(1), solution number(1)

Maturity scale raw canonical shape-checked rule-derived certified

Inbound mentions (6)

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.

alternativeToAlternative to(1)

demonstratesDemonstrates(1)

identifiesSolutionIdentifies Solution(1)

proposesSolutionProposes Solution(1)

recommendsRecommends(1)

requiresLargerModelRequires Larger Model(1)

Other facts (18)

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.

18 facts
PredicateValueRef
Rdf:typeProposed Solution[2]
Rdf:typeSolution[3]
Rdf:typeProposed Solution[4]
Relies on Factual Recall ScalingParameter Count[1]
Solution Number2[2]
Involves ActionDisable Vercel Github Checks[2]
Location of ActionVercel Dashboard[2]
Agent Capable of ActionAjaxdavis[2]
Proposes Value2048[3]
Has Context Size2000[4]
Ordinal Position2[5]
Proposes ActionLower LR[5]
Alternative toSolution 1[6]
Allows Vector Dimension3[6]
Requires Collection Dimension3[6]
Methodcollection-adjustment[6]
Solves ProblemDimension Mismatch Problem[6]
Has Code ExampleCorrected Code Solution 2[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.

reliesOnFactualRecallScalingblah/watt-activation/part-178
ex:parameter-count
typeblah/blah/11
ex:ProposedSolution
solutionNumberblah/blah/11
2
involvesActionblah/blah/11
ex:disable-vercel-github-checks
locationOfActionblah/blah/11
ex:vercel-dashboard
agentCapableOfActionblah/blah/11
ex:ajaxdavis
typeblah/training-and-evals/11
ex:Solution
proposesValueblah/training-and-evals/11
2048
typeblah/watt-activation/26
ex:ProposedSolution
hasContextSizeblah/watt-activation/26
2000
ordinalPositionblah/watt-activation/44
2
proposesActionblah/watt-activation/44
Lower LR
titlebeam/b99b8773-86e1-4542-99be-ea39973cacf9
Adjust the Collection Dimension
alternativeTobeam/b99b8773-86e1-4542-99be-ea39973cacf9
ex:solution-1
allowsVectorDimensionbeam/b99b8773-86e1-4542-99be-ea39973cacf9
3
requiresCollectionDimensionbeam/b99b8773-86e1-4542-99be-ea39973cacf9
3
methodbeam/b99b8773-86e1-4542-99be-ea39973cacf9
collection-adjustment
solvesProblembeam/b99b8773-86e1-4542-99be-ea39973cacf9
ex:dimension-mismatch-problem
hasCodeExamplebeam/b99b8773-86e1-4542-99be-ea39973cacf9
ex:corrected-code-solution-2

References (6)

6 references
  1. [1]Part 1781 fact
    ctx:discord/blah/watt-activation/part-178
  2. [2]115 facts
    ctx:discord/blah/blah/11
    • full textblah-11
      text/plain3 KBdoc:agent/blah-11/200680a7-8423-48ff-a9a2-261cde3d3bc6
      Show excerpt
      [2026-02-21 17:29] foxhop.: The annoying errors are almost certainly from Vercel's GitHub integration which creates check runs on every commit. This is controlled by Vercel's project settings, not GitHub Actions.
  3. [3]112 facts
    ctx:discord/blah/training-and-evals/11
    • full texttraining-and-evals-11
      text/plain3 KBdoc:agent/training-and-evals-11/5e6024b9-dce0-4ec3-b112-06d13e1c5c96
      Show excerpt
      [2026-02-21 16:35] ajaxdavis: ``` ● The models are all up — the problem is the eval runner itself. Here's what's happening:
  4. [4]262 facts
    ctx:discord/blah/watt-activation/26
    • full textwatt-activation-26
      text/plain3 KBdoc:agent/watt-activation-26/33435d36-2781-4f3b-ab35-e0cdf2197a23
      Show excerpt
      [2026-03-06 16:23] xenonfun: On making conversational: ``` Right — no EOS token means the model just generates forever until you hit max_new_tokens. A few paths to fix this, in order of effort: 1. Heuristic stop (zero retraining, works
  5. [5]442 facts
    ctx:discord/blah/watt-activation/44
    • full textwatt-activation-44
      text/plain3 KBdoc:agent/watt-activation-44/35af5294-4efe-43b8-a39d-23df23087144
      Show excerpt
      [2026-03-07 05:32] xenonfun: ``` Not good. The loss trajectory tells a clear story: ┌────────┬─────────────────────────┐ │ Iters │ Avg Loss │ ├────────┼─────────────────────────┤ │ 0-2K │ 6.56 (learning) │
  6. ctx:claims/beam/b99b8773-86e1-4542-99be-ea39973cacf9
    • full textbeam-chunk
      text/plain1 KBdoc:beam/b99b8773-86e1-4542-99be-ea39973cacf9
      Show excerpt
      If you want to keep the collection dimension at 128, you need to adjust the vectors to have 128 dimensions each. For example: ```python vectors = [ [1.0] * 128, # A vector with 128 elements, all initialized to 1.0 [2.0] * 128 # A

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.