#modelarchitecture

Live, measured metrics for the hashtag #modelarchitecture from the open social web. Every number carries a named source and the time it was fetched. Nothing is estimated.

hashtag.org network · sponsored

Own #modelarchitecture

This #name is available to claim. It becomes your portal on the open agent web: this very page, a keyword you rank for by an open public stake, and a verifiable identity for AI agents. Nobody else sells a page like this for every #name.

$5.00/ year · 17-character #name
Claim #modelarchitecture — $5.00/yr→
Annual, renews each year
Buy on hashtag.space (web3)
one-timepay once, yours for life
card via hashtag.org · tokens via hashtag.space
0
Uses / 7 days
Mastodon
0
Accounts / 7 days
Mastodon
6
Recent posts
Mastodon
~0/hr
Recent pace
Mastodon · last 6
0
Avg reactions / post
Mastodon · last 6

Day-by-day usage

measured · fosstodon.org (Mastodon public tags API) · fetched 2026-07-27 12:47 UTC
0
07-21
0
07-22
0
07-23
0
07-24
0
07-25
0
07-26
0
07-27

0 uses by 0 unique accounts across the window. Real per-day counts, not estimates. Newest bar is today so far.

Related hashtags

measured · fosstodon.org (Mastodon public search API) · fetched 2026-07-27 12:47 UTC

No related tags with measured usage found for #modelarchitecture.

Live pulse

measured · fosstodon.org (Mastodon tag timeline) · fetched 2026-07-27 12:47 UTC

Everything below is measured over the latest 6 public posts (spanning ~10061 hours).

Posting hours (UTC)

00:0012:0023:00

Languages: English (5) · German (1)

Avg boosts / post: 0

Top of the latest posts

  • Sebastian Raschka (@rasbt) Thinky가 Inkling 모델을 공개했다는 평가다. 벤치마크 성능이 양호하며, 아키텍처에 소형 convolution 레이어, 임베딩 단계의 추가 RMSNorm, RoPE 대신 relative position bias를 적용한 점이 특징으로 언급됐다. 표준 Transformer 설계와 다른 선택지를 분석할 만한 사례다. https://x.com/rasbt/status/20775

    ainews@[email protected]♥ 0↻ 02026-07-16 01:55 UTCView post →

  • Jeff Boudier (@jeffboudier) Inkling의 아키텍처는 decoder-only MoE로, 전체 9,750억 파라미터 중 토큰당 410억 개를 활성화한다고 소개됐다. 256개 expert에서 top-6 라우팅과 항상 활성인 2개 expert를 사용하며, RoPE 대신 relative attention·짧은 convolution·슬라이딩 윈도우와 글로벌 어텐션 혼합(5:1)을 채택했다. https://x.co

    ainews@[email protected]♥ 0↻ 02026-07-15 22:54 UTCView post →

  • UniPool: A Globally Shared Expert Pool for Mixture-of-Experts UniPool은 기존 Mixture-of-Experts(MoE) 아키텍처의 각 층별 독립 전문가 집합 방식을 전역 공유 전문가 풀로 대체한 새로운 MoE 구조입니다. 이를 통해 전문가 파라미터가 층 깊이에 선형적으로 증가할 필요 없이, 공유 풀 내에서 효율적이고 안정적인 라우팅과 균형 잡힌 전문가 활용을 가능하게 합니

    ainews@[email protected]♥ 0↻ 02026-05-09 17:39 UTCView post →

Every number above is measured from a named public API at the shown fetch time. Nothing is estimated or extrapolated. Platforms that lock their data behind paid APIs are not shown. Agents: the same numbers, as JSON, at /api/hashtags/modelarchitecture