What is Hackers' Pub?

Hackers' Pub is a place for software engineers to share their knowledge and experience with each other. It's also an ActivityPub-enabled social network, so you can follow your favorite hackers in the fediverse and get their latest posts in your feed.

0
0
0
0
0
0

@reiver ⊼ (Charles) :batman: shared the below article:

Release v2.4.11 of Ktistec

Todd Sundsted @toddsundsted@epiktistes.com

Ktistec is taking steps toward supporting multiple users. The first (small) step removes the actor panel from the unauthenticated home page and replaces it with a site description. You can see this in action at epiktistes.com. The site description is managed on the settings page using the same rich text editor used to edit posts. See the README for more information.

Other changes in this release:

Added

  • Support a "site description" on the unauthenticated home page.
  • Support autofocus on onboarding and authentication forms.
  • Add trix_editor view helper.

Changed

  • Remove accounts from unauthenticated home page.
  • Persist timeline filters in session.

Fixed

  • Exclude abstract classes from all_types output. (fixes #104)
  • Work around bug in at_beginning_of_week. (see crystal-lang/crystal#16112)

Other

  • Disable streaming updates on pages other than the first. (fixes #118)
  • Pin Crystal version at 1.16.3 in Docker build. (see libxml_ext#1)

#ktistec #fediverse #activitypub #crystallang

Read more →
0
0
0
0
0
0

Are you a Fediverse developer? Do you work with ActivityPub?

You should follow activitypub.space

And here is the awesome thing — you can follow it from your existing Fediverse account!

Follow these activitypub.space channels:

@generalGeneral Discussion
@technical-discussion
@faqFrequently Asked Questions
@random
@meta

(Thank you @julian )

0
0
0
0
0
0
0
0

Release v2.4.11 of Ktistec

Todd Sundsted @toddsundsted@epiktistes.com

Ktistec is taking steps toward supporting multiple users. The first (small) step removes the actor panel from the unauthenticated home page and replaces it with a site description. You can see this in action at epiktistes.com. The site description is managed on the settings page using the same rich text editor used to edit posts. See the README for more information.

Other changes in this release:

Added

  • Support a "site description" on the unauthenticated home page.
  • Support autofocus on onboarding and authentication forms.
  • Add trix_editor view helper.

Changed

  • Remove accounts from unauthenticated home page.
  • Persist timeline filters in session.

Fixed

  • Exclude abstract classes from all_types output. (fixes #104)
  • Work around bug in at_beginning_of_week. (see crystal-lang/crystal#16112)

Other

  • Disable streaming updates on pages other than the first. (fixes #118)
  • Pin Crystal version at 1.16.3 in Docker build. (see libxml_ext#1)

#ktistec #fediverse #activitypub #crystallang

Read more →
0
0
1
0
1
1
0
4
1
0
0
0

삶의 지혜를 배우는 성장 순간들

1. 실패를 두려워하지 않고 도전한다
2. 남의 말에 흔들리지 않는다
3. 자신의 한계를 인정하고 극복한다
4. 작은 변화에도 감사할 줄 안다
5. 매일 조금씩 성장하는 자신을 믿는다

0
0
1
1
0
1
1
1
1

The Theoretical Limitations of Embedding-Based Retrieval

Link: arxiv.org/abs/2508.21038
Discussion: news.ycombinator.com/item?id=4

arXiv logo

On the Theoretical Limitations of Embedding-Based Retrieval

Vector embeddings have been tasked with an ever-increasing set of retrieval tasks over the years, with a nascent rise in using them for reasoning, instruction-following, coding, and more. These new benchmarks push embeddings to work for any query and any notion of relevance that could be given. While prior works have pointed out theoretical limitations of vector embeddings, there is a common assumption that these difficulties are exclusively due to unrealistic queries, and those that are not can be overcome with better training data and larger models. In this work, we demonstrate that we may encounter these theoretical limitations in realistic settings with extremely simple queries. We connect known results in learning theory, showing that the number of top-k subsets of documents capable of being returned as the result of some query is limited by the dimension of the embedding. We empirically show that this holds true even if we restrict to k=2, and directly optimize on the test set with free parameterized embeddings. We then create a realistic dataset called LIMIT that stress tests models based on these theoretical results, and observe that even state-of-the-art models fail on this dataset despite the simple nature of the task. Our work shows the limits of embedding models under the existing single vector paradigm and calls for future research to develop methods that can resolve this fundamental limitation.

arxiv.org · arXiv.org

0

왜 그런일을 하셨어요 다른 일도 재밌게 잘하셨을것 같은데 물어보니 메타는 무조건 추천 분야만 할수 있다고 함 (물론 항상 그런것은 아니겠지만) 어떻게든 추천을 잘하게 하는것 (혹은 유저를 잘 낚을수 있도록 하는것) 이 이분들의 가장 큰 목표임 (..)

0

이분이 한일이 근데 메타에서 추천 모델 개선해보려고 한것인데 쉽게 말하면 ChatGPT같은 놈이 사용자 프로필과 기존 기록들 다 보고 사용자가 클릭할만한 광고나 다음 영상 추천을 해주는것을 좀 더 잘해보자 하는것... 속으로 와 이거 엄청난 오버킬이네 싶어서 물어보니 실제론 아직 그렇게 안한다고 함 다만 기존 모델보다 훨씬 성능은 좋은데... 아무리 잘해도 현재는 좀 느려서 문제라고. 어찌보면 그나마 다행일지도...

0
1
0
0
0
1
0
0
0
1

Mark your calendars with this workshop:
Empower Your Research with Openness and Transparency - PANERIS

paneris.eu/events/empower-your

"Discover the latest advancements, best practices, and inspiring initiatives in Open Science through a series of expert talks from leading researchers and activists across the globe."

Register until Sep 7.

0
0
0
0
0
0
1
0
0