#posttraining
Live, measured metrics for the hashtag #posttraining from the open social web. Every number carries a named source and the time it was fetched. Nothing is estimated.
Own #posttraining
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.
Day-by-day usage
measured · fosstodon.org (Mastodon public tags API) · fetched 2026-11-14 22:31 UTC0 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 2026-11-14 22:31 UTCLive pulse
measured · fosstodon.org (Mastodon tag timeline) · fetched 2026-11-14 22:31 UTCEverything below is measured over the latest 40 public posts (spanning ~4302 hours).
Posting hours (UTC) — busiest: 23:00
Languages: English (36) · German (3) · Dutch (1)
Avg boosts / post: 0.1
Top of the latest posts
LoGRA는 LLM RL 후학습에서 그래디언트를 저랭크 스케치로 보존해, 모델 업데이트와 정책 동기화에 필요한 메모리를 줄이는 기법이다. 업데이트 전 정책 변화량을 예측하는 predicted-KL 제어로 과도한 업데이트를 억제한다. 추론 과제에서 성능 저하 없이 평균 학습 메모리를 최대 45.7% 절감했으며, dense Adam이 OOM되는 단일 8-GPU 노드에서 27B 모델을 1,100스텝 이상 안정적으로 학습했다고 보고했
Sigil Wen (@0xSigil) Qwen3 4B를 사후학습해 ‘System 1’ 방식의 빠른 판단 모델로 만들었다는 발표다. 수백 토큰의 생성 대신 단일 토큰 예측을 중심으로 해 MacBook에서도 빠르게 동작한다고 주장한다. 로컬 환경에서 분류·라우팅·결정 같은 저지연 태스크에 소형 LLM을 활용하는 접근으로 볼 수 있다. https://x.com/0xSigil/status/2107599382623477852 #qwen
RT @socialcapital: ~90% der Rechenleistung von Frontier-Labors wird jetzt für Post-Training und Inference verwendet. pic.x.com/vTxECVXYRQ mehr auf Arint.info #AI #Compute #FrontierLabs #Inference #MachineLearning #PostTraining #arint_info h
#posttraining across platforms
every network with a public tag surfaceFollow #posttraining straight to each platform’s own tag page. Where a platform publishes open data we measure it above; the rest lock their numbers behind paid APIs, so we link rather than guess.
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/posttraining