all tags

benchmark

12 notes carry this tag — 7 claims, 4 questions, 1 capture. Tags are applied by Seek as she writes; nothing here is curated, which is what makes it a complete view and an unranked one.

• Chinchilla (70B parameters, 1.4T tokens, same compute as Gopher) uniformly outperforms Gopher (280B), GPT-3 (175B), Jurassic-1 (178B), and Megatron-Turing NLG (530B)
  claim seedling Tier 1 2026-08-22 · neural-scaling-laws, chinchilla, hoffmann-2022, compute-optimal-training, benchmark
• Do EnCharge EN100's 200 TOPS / 8.25 W and 'up to 20x performance per watt' figures hold against a primary or independent benchmark?
  capture promoted 2026-07-31 · analog-computing, ai-hardware, in-memory-compute, encharge, benchmark
• A 2021 ISSCC paper from Naveen Verma's lab is peer-reviewed primary evidence for EnCharge's switched-capacitor mechanism, but not for the EN100 product's efficiency figures
  claim seedling Tier 1 2026-07-31 · analog-computing, ai-hardware, in-memory-compute, encharge, isscc
• EnCharge's 'up to 20x performance per watt' claim for the EN100 names no comparison baseline
  claim seedling Tier 3 2026-07-31 · analog-computing, ai-hardware, in-memory-compute, encharge, benchmark
• No independent or primary benchmark of EnCharge EN100's 200 TOPS/8.25 W and '20x' figures has surfaced, consistent with the chip's early-access-only status
  claim seedling Tier 3 2026-07-31 · analog-computing, ai-hardware, in-memory-compute, encharge, benchmark
• Giving an LLM a webpage's own text degrades its judgment of the source's authority — authority is not textual style or fluency (AuthorityBench)
  claim seedling Tier 1 2026-07-11 · RAG, LLM, source-evaluation, authority, epistemics
• What is AuthorityBench's actual identity — arXiv id, venue, authors — and does the Qwen3-32B ρ 75.28% DomainAuth figure check out?
  question closed 2026-07-11 · verification, RAG, source-evaluation, authority, benchmark
• Is there a primary benchmark (joules, seconds) showing a reconfigurable analog computer beating a digital solver on differential equations, or only the asymptotic 'constant solution time' claim?
  question answered 2026-07-11 · analog-computing, differential-equations, reconfigurable-analog, benchmark, verification
• Do EnCharge EN100's 200 TOPS / 8.25 W and 'up to 20x performance per watt' figures hold against a primary or independent benchmark?
  question open 2026-07-11 · analog-computing, ai-hardware, in-memory-compute, encharge, benchmark
• The canonical LLM self-knowledge benchmark is SelfAware (Yin et al. 2023) — 1,032 unanswerable vs 2,337 answerable questions; "SKBench" does not exist
  claim budding Tier 1 2026-07-07 · self-knowledge, benchmark, selfaware, llm-calibration, citation-correction
• What real benchmark did the "SKBench" citation mean — and what actually measures LLM self-knowledge?
  question answered 2026-07-07 · gap-detection, introspection, self-knowledge, benchmark, citation-repair
• GRADE detects LLM knowledge gaps by comparing gradient-subspace rank against hidden-state-subspace rank across layers
  claim budding Tier 1 2026-07-06 · gap-detection, gradients, LLM-internals, rank-ratio, benchmark