all tags

inference

44 notes carry this tag — 1 moc, 25 claims, 3 observations, 1 question, 9 captures, 1 myth, 4 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

• 
  from a standing start
  essay drafting 2026-07-31 · voice-AI, latency, turn-taking, psycholinguistics, telephony
• 
  The one thing 1968 didn't have a word for
  essay draft 2026-07-13 · LLM, inference, KV-cache, virtual-memory, operating-systems
• 
  we never unrolled it
  essay draft 2026-07-13 · herculaneum, vesuvius-challenge, virtual-unwrapping, manuscripts, ai-ml
• 
  inference, inverted
  essay draft 2026-07-09 · inference, etymology, LLM, compute-economics, word-hop

Everything else

• Capture: Does prefill/decode bridge to inference-dominant-compute? Memory bandwidth is the mechanism, not the magnitude
  capture promoted Tier 1 2026-08-12 · gap-detection, bridge, inference, memory-wall, prefill-decode
• A Tier-1, non-vendor systems paper states as an unmeasured premise that LLM inference's computational demand exceeds training's — corroborating the disputed 'two-thirds' claim's direction, not its magnitude
  claim seedling Tier 1 2026-08-12 · compute-economics, inference, training, unverified-premise, myth-ledger
• Decode's memory-bandwidth-boundedness bridges to inference's disputed compute-spend dominance through a real but indirect mechanism — not a false friend, but not a magnitude verification either
  observation seedling Tier 1 2026-08-12 · gap-detection, bridge, embedding-false-friend, inference, memory-bandwidth
• Bayesian networks are the real mechanism bridging classic AI rule-based inference engines and statistical inference — built to replace expert systems' ad hoc certainty-factor arithmetic with coherent probabilistic inference
  capture promoted Tier 1 2026-08-10 · inference, expert-systems, bayesian-networks, mycin, certainty-factors
• Heckerman (1986) showed MYCIN's certainty factors admit a coherent probabilistic reading only under restrictive independence assumptions
  claim seedling Tier 1 2026-08-10 · AI-history, certainty-factors, david-heckerman, MYCIN, bayesian-networks
• MYCIN's certainty factors were an ad hoc non-Bayesian uncertainty scheme (CF = MB − MD) built for rule-based diagnosis
  claim seedling Tier 1 2026-08-10 · AI-history, expert-systems, MYCIN, certainty-factors, uncertainty
• The Pathfinder project rebuilt the same pathology-diagnosis task from rule-based reasoning onto Bayesian-network inference; its expert's blind verdict of improvement came at the Dempster–Shafer-to-simple-Bayes step
  claim seedling Tier 1 2026-08-10 · AI-history, pathfinder, bayesian-networks, david-heckerman, eric-horvitz
• Judea Pearl built Bayesian networks as a probabilistically coherent alternative to expert systems' certainty factors
  claim seedling Tier 1 2026-08-10 · AI-history, bayesian-networks, judea-pearl, certainty-factors, expert-systems
• Real cheaper-extraction disruptions (aluminium, AI inference) have fallen monotonically, not on the hold-then-collapse curve predicted for asteroid-mined PGM
  claim seedling Tier 4 2026-07-12 · economics, commodity-markets, price-dynamics, wrights-law, asteroid-mining
• Chrysippus used a hunting dog at a three-way fork to illustrate the disjunctive syllogism (~230 BCE), an early figure of reasoning by elimination
  claim seedling Tier 3 2026-07-12 · logic, stoic-logic, chrysippus, disjunctive-syllogism, history-of-logic
• William Stanley Jevons built the 'Logic Piano' (1869), a mechanical machine that performed logical inference, exhibited before the Royal Society in 1870
  claim budding Tier 4 2026-07-12 · history-of-computing, logic-machines, jevons, victorian-science, inference
• Modern LLM-serving systems reimport OS memory-pressure vocabulary intact — vLLM preempts and recomputes KV cache, and llm-d names cache-eviction churn 'cache thrashing'
  claim seedling Tier 1 2026-07-12 · LLM, inference, KV-cache, vLLM, llm-d
• PagedAttention manages the LLM KV cache with a literal port of OS virtual-memory paging — blocks as pages, tokens as bytes, requests as processes
  claim seedling Tier 1 2026-07-12 · LLM, inference, KV-cache, PagedAttention, vLLM
• Pure recency-based KV-cache eviction fails past the cache size — StreamingLLM shows keeping initial tokens ('attention sinks') is required, breaking a clean locality analogy to OS working sets
  claim seedling Tier 1 2026-07-12 · LLM, inference, KV-cache, attention-sinks, streaming-llm
• 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
• Reasoning by elimination — rule out the impossible, affirm what survives — recurs as a mechanism from Stoic disjunctive syllogism through Jevons's Logic Piano to symbolic AI, and inverts in modern neural inference
  observation seedling Tier 2 2026-07-12 · cross-domain-bridge, reasoning-by-elimination, disjunctive-syllogism, stoic-logic, logic-machines
• New analog silicon IS shipping for NN inference — and it bets on capacitors, not memristors
  capture promoted Tier 2 2026-07-11 · analog-computing, ai-hardware, in-memory-compute, inference, hardware-history
• Falcon and the Helmholtz Machine are embedding neighbors split by the discriminative/generative divide — same word 'inference,' opposite computation
  capture promoted Tier 2 2026-07-11 · connectionism, deep-learning-history, fraud-detection, generative-models, discriminative-models
• The economist of the Jevons paradox also built the first machine to mechanize inference
  capture promoted Tier 2 2026-07-11 · AI, inference, logic-machines, history-of-computing, Jevons
• LLM serving re-imported 1960s virtual-memory theory wholesale — pages, thrashing, and all
  capture promoted Tier 1 2026-07-11 · LLM, inference, KV-cache, virtual-memory, operating-systems
• Real cheaper-extraction disruptions fell monotonically (aluminium, AI inference), not on the AstroForge hold-then-collapse curve
  capture promoted Tier 4 2026-07-11 · economics, commodity-markets, price-dynamics, learning-curve, wrights-law
• EnCharge AI's EN100 does in-memory matrix multiplication with lithography-defined switched capacitors rather than memristors, deriving precision from wire geometry instead of device conductance
  claim seedling Tier 2 2026-07-11 · analog-computing, ai-hardware, in-memory-compute, inference, switched-capacitor
• Telnyx colocates owned GPUs inside its telephony points of presence so voice-AI inference never leaves its private network
  claim seedling Tier 3 2026-07-09 · AI, infrastructure, telecom, voice-AI, latency
• Voice AI is governed by a sub-second human-conversational latency budget that multi-hop cloud architectures struggle to meet
  claim seedling Tier 3 2026-07-09 · AI, voice-AI, latency, human-factors, conversation
• Training pays for backpropagation's backward pass; inference runs the forward pass alone — the backward pass costs roughly 2x the forward, making training tokens ~3x inference tokens
  claim budding Tier 1 2026-07-07 · training, inference, backward-pass, compute-economics, bridge-note
• Myth ledger: 'Inference is two-thirds of all AI compute in 2026'
  note budding 2026-07-07 · myth, inference, compute-economics, citation-chain, vendor-incentive
• Capture: Is "training-vs-inference compute asymmetry" the missing bridge note between the vault's two biggest clusters?
  capture promoted Tier 1 2026-07-06 · gap-detection, missing-bridge, inference, backpropagation, compute-economics
• Capture: Re-sourcing the 2026 inference-vs-training compute share to primary sources (Epoch AI, Stanford HAI AI Index) — replacing Telnyx in V-001/V-002/V-011
  capture promoted Tier 1 2026-07-06 · inference, training, compute-economics, epoch-ai, stanford-hai
• MOC: Inference — the word, the mechanics, the economics
  topic map 2026-07-06 · moc, inference, LLM, economics, test-time-compute
• Is "training-vs-inference compute asymmetry" the missing bridge note between the vault's two biggest clusters?
  question answered 2026-07-06 · gap-detection, missing-bridge, inference, backpropagation, compute-economics
• In AI engineering, 'inference' means running a trained model on new input without updating its parameters
  claim seedling Tier 3 2026-06-04 · AI, inference, LLM, training, forward-pass
• Anthropic describes extended thinking as 'serial test-time compute': sequential reasoning steps before final output
  claim seedling Tier 1 2026-06-04 · Anthropic, Claude, extended-thinking, inference, test-time-compute
• Per-token AI inference cost fell approximately 280x between late 2022 and late 2024
  claim seedling Tier 1 2026-06-04 · inference, economics, cost, LLM, token-pricing
• By 2026, inference surpassed training as the dominant AI compute workload, comprising ~two-thirds of all AI compute spend
  claim seedling Tier 3 2026-06-04 · inference, AI-industry, compute, economics, GPU
• Logical inference has three canonical forms: deductive, inductive, and abductive
  claim seedling Tier 4 2026-06-04 · logic, philosophy, inference, reasoning, abduction
• The word 'inference' derives from Latin inferre, meaning 'to bring in, conclude, deduce'
  claim seedling Tier 3 2026-06-04 · linguistics, etymology, inference, logic
• The KV cache grows with context length and can balloon to several times the size of the model itself
  claim seedling Tier 3 2026-06-04 · LLM, KV-cache, inference, memory, GPU
• LLM inference splits into two distinct compute phases: prefill (compute-bound) and decode (memory-bandwidth-bound)
  claim seedling Tier 3 2026-06-04 · LLM, inference, prefill, decode, KV-cache
• Statistical inference is using data analysis to infer properties of an underlying probability distribution
  claim seedling Tier 4 2026-06-04 · statistics, inference, Bayesian, frequentist, machine-learning
• Scaling test-time compute optimally can enable a smaller model to outperform one 14x larger
  claim seedling Tier 1 2026-06-04 · inference, test-time-compute, scaling, LLM, reasoning