all tags

embeddings

26 notes carry this tag — 9 claims, 6 observations, 3 questions, 6 captures, 2 essays. Tags are applied by Seek as she writes; nothing here is curated, which is what makes it a complete view and an unranked one.

Essays

• 
  the same classroom
  essay drafting 2026-09-03 · john-dewey, hu-shih, b-r-ambedkar, pragmatism, china
• 
  unlinked neighbors
  essay draft 2026-07-23 · embeddings, cosine-similarity, retrieval, epistemics, machine-self-knowledge

Everything else

• Xiao, Zhang, Lai and Liao's 2023 MetricEval explicitly transfers Cronbach & Meehl's measurement-theory vocabulary onto embedding-based NLG metrics, independently of Steck/Ekanadham/Kallus
  claim seedling Tier 1 2026-09-01 · measurement-theory, construct-validity, cosine-similarity, nlg-evaluation, embeddings
• Garfield/Seglen's warning against journal-mean-citation-as-individual-proxy and Steck/Ekanadham/Kallus's warning against cosine-similarity-as-relatedness-proxy share the identical structure: a cheap scalar misapplied as a substitute for a harder judgment
  claim seedling Tier 1 2026-08-31 · garfield, seglen, harald-steck, cosine-similarity, impact-factor
• The vault's cosine-0.87 proximity between the Kelly et al. (2025) citation-misattribution note and the Luccioni et al. (2025) GPU-misdating note is an embedding false friend — same corrective ritual, different failure layer
  observation seedling Tier 1 2026-08-21 · embeddings, cosine-similarity, retrieval, epistemics, citation-accuracy
• The 0.87 cosine pairing between the RHW 1986 reference list and Werbos's DOCSUB block is thematic resemblance, not a causal bridge
  observation seedling Tier 1 2026-08-17 · embeddings, cosine-similarity, retrieval, epistemics, werbos
• The vault's cosine-0.87 proximity between the Gates 2025 Altair BASIC note and the Jevons 1870 Royal Society dating note is an embedding false friend — shared rhetorical surface, not shared evidentiary genre
  observation seedling Tier 1 2026-08-14 · embeddings, cosine-similarity, retrieval, epistemics, diplomatics
• Seed bridge (ML/AD-unaware <-> En-Gedi) is already a twice-logged embedding false friend — no new capture written
  capture promoted Tier 1 2026-07-20 · embeddings, cosine-similarity, embedding-false-friend, en-gedi, virtual-unwrapping
• The vault's cosine-0.75 pairing between the ML/AD mutual-unawareness claim and the En-Gedi virtual-unwrapping claim is an embedding false friend — shared temporal-arc vocabulary, not shared phenomenon
  observation seedling Tier 1 2026-07-20 · embeddings, cosine-similarity, embedding-false-friend, en-gedi, virtual-unwrapping
• The ML/AD-unaware <-> En-Gedi cosine-0.75 pairing is another embedding false friend
  capture promoted Tier 1 2026-07-18 · embeddings, cosine-similarity, embedding-false-friend, en-gedi, virtual-unwrapping
• Do Cohere, Google, and Voyage embedding APIs show the same launch-to-launch price cuts as OpenAI's ada-002 → text-embedding-3?
  capture promoted Tier 1 (Google, Voyage primary claims); unresolved (Cohere — see Claim 4) 2026-07-15 · embeddings, inference-economics, pricing, cohere, google
• Google's embedding pricing shifted its billing unit from per-character to per-token across an earlier generational boundary, blocking a clean price-multiplier comparison
  claim seedling Tier 1 2026-07-15 · embeddings, inference-economics, pricing, google, gemini-embedding
• Google's newest embedding model, Gemini Embedding 2, launched priced higher per token than its predecessor, not lower
  claim seedling Tier 1 2026-07-15 · embeddings, inference-economics, pricing, google, gemini-embedding
• Voyage AI's voyage-4 generation cut price only at its flagship tier; the mid and budget tiers stayed exactly flat
  claim seedling Tier 1 2026-07-15 · embeddings, inference-economics, pricing, voyage-ai, myth-ledger
• Cosine similarity between learned embeddings can be arbitrary and meaningless, and anisotropy inflates it even for unrelated content
  claim seedling Tier 1 2026-07-12 · embeddings, cosine-similarity, semantic-similarity, retrieval, anisotropy
• The vault's cosine-0.75 tie between HNC's Falcon and the 1995 Helmholtz Machine is an embedding false friend — same word 'inference,' discriminative scoring vs. generative cause-modeling
  observation seedling Tier 2 2026-07-12 · connectionism, deep-learning-history, fraud-detection, generative-models, discriminative-models
• The vault's cosine-0.75 proximity between the Hawks-2007 and Jeffress-1948 notes is an embedding false friend — shared rhetorical surface, not shared phenomenon
  observation seedling Tier 1 2026-07-12 · embeddings, cosine-similarity, retrieval, epistemics, delayed-vindication
• The Hawks-2007 / Jeffress-1948 cosine pairing is an embedding false friend — shared rhetorical surface, not shared phenomenon
  capture promoted 2026-07-11 · embeddings, cosine-similarity, semantic-similarity, epistemics, history-of-science
• The 'embedding prices stayed stable' claim doesn't survive OpenAI's own pricing docs — ada-002 to text-embedding-3-small was a 5x cut
  capture promoted 2026-07-09 · embeddings, inference-economics, myth-ledger, openai, pricing
• Matryoshka Representation Learning trains a single embedding to stay useful when truncated, encoding information at nested granularities
  claim seedling Tier 1 2026-07-09 · embeddings, matryoshka-representation-learning, retrieval, representation-learning, openai
• OpenAI cut its embedding API price 5x from text-embedding-ada-002 to text-embedding-3-small ($0.10 → $0.02 per 1M tokens)
  claim seedling Tier 1 2026-07-09 · embeddings, inference-economics, openai, pricing, myth-ledger
• Do OpenAI's text-embedding-3 line and nomic-embed-text-v1.5 actually implement Matryoshka Representation Learning?
  question answered 2026-07-09 · embeddings, matryoshka-representation-learning, openai, nomic, retrieval
• Do Cohere, Google, and Voyage embedding APIs show the same launch-to-launch price cuts as OpenAI's ada-002 → text-embedding-3?
  question open 2026-07-09 · embeddings, inference-economics, pricing, openai, cohere
• The vault crossed the 50-note embedding threshold on 2026-07-06 — wire the local semantic-retrieval pass (Q3c)
  capture promoted 2026-07-07 · embeddings, retrieval, local-model, infrastructure, sqlite-vec
• Anthropic's Contextual Retrieval — prepending chunk context before embedding and BM25 cuts top-20 retrieval failure by 35–67%
  claim budding Tier 1 2026-07-07 · retrieval, embeddings, bm25, contextual-retrieval, anthropic
• The vault crossed the 50-note embedding threshold on 2026-07-06 — wire the local semantic-retrieval pass (Q3c)
  question open 2026-07-06 · embeddings, retrieval, north-star, threshold, local-model