mstdn.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A general-purpose Mastodon server with a 500 character limit. All languages are welcome.

Administered by:

Server stats:

0
active users

#huggingface

1 post1 participant1 post today

I saw this yesterday. I'm wondering how hard it would be to pair the Mini Lite with a @homeassistant back end and let #HomeAssistant use it for sensors and as a satellite for voice commands. The article says they're "fully programmable in Python", so it FEELS like that should be a possibility, but I don't have the skill set that would be required to do that. I might buy one if I thought I could use one that way.

techcrunch.com/2025/07/09/hugg

TechCrunch · Hugging Face opens up orders for its Reachy Mini desktop robots | TechCrunchHugging Face is releasing two versions of its desktop robot meant for designing and coding AI applications.

Hugging Face’s Reachy Mini is an open source robot aimed at developers, priced at $299 and up

AI software company Hugging Face is now taking pre-orders for a small robot aimed at developers an folks learning to code. It’s called Reachy Mini, and it’s basically a programmable toy that sits on your desk and interacts with people around it by moving its head, rotating its body, or waving its antenna.

The robot comes in two versions: there’s a Reachy Mini Lite that costs $299 and which […]

#ai #huggingFace #reachy #reachyMini #robots

Read more: liliputing.com/hugging-faces-r

📢 Thrilled to share that, through a collaborative effort between the @OpenSearchProject and @huggingface neural sparse models are now available in the Sentence Transformers library. 🤗

The Sentence Transformers (a.k.a. SBERT) library, developed by @UKPLab and maintained by #HuggingFace, is a Python framework designed to generate semantically meaningful embeddings for sentences, paragraphs, and images.

opensearch.org/blog/neural-spa

Congrats to all involved! 👏

Собираем MVP product search: дообучение E5 и веб-сервис для сравнения поисквых выдач

Что важнее: создать продукт , или доставить его до пользователя ? Оба этапа необходимы. Сегодня обсудим второй . Как нам построить поисковую e-com систему. Покажем, что в слово логистика товара входят сложные задачи не только: перевезти наушники из Китая в Америку , но и настройка поисковой выдачи по запросу. Быстро соберем поисковой MVP-сервис . Дообучим модель E5 на реальных данных от Amazon . Определим метрики качества и сравним BM25 , pretrain E5 и fine-tune E5 . Так же взглянем глазами с отладочной информацией и проанализируем изменения поисковых выдач . И под конец обсудим каких технологий еще не хватает и можно добавить, если возникают соответствующие трудности. Погрузиться в семантический поиск →

habr.com/ru/companies/datafeel

ХабрСобираем MVP product search: дообучение E5 и веб-сервис для сравнения поисквых выдачЧто важнее: создать продукт , или доставить его до пользователя ? Оба этапа необходимы. Сегодня обсудим второй . Как нам построить поисковую e-com систему. Покажем, что в слово логистика товара входят...
Continued thread

(Linux news in previous posts of thread)

FOSS NEWS

Odysee presents new AR-based payment system, discontinues Stripe payment and LBC content boosting:
odysee.com/@Odysee:8/introduci
(Discontinuing LBC support probably means the platform won't be connected to the LBRY blockchain anymore. Also, content discoverability will be hurt pretty much, even if creators can get monetized, smaller creators will have a much harder time being discovered. I get that their new parent company Arweave is probably forcing them to switch to their crypto stuff and ditch LBRY, but it will probably just hurt the platform in the long run, especially because the new payment system requires signing up to an additional service, which many people are not a big fan of.)

Thunderbird 140 released with "Mark as Spam" and "Mark as Starred" actions for notifications, reorganized settings, bug fixes:
9to5linux.com/thunderbird-140-

ONLYOFFICE Documents 9.0 for Android released with OCR support, DocSpace version history, Markdown and OpenDocument Graphics support, etc.:
alternativeto.net/news/2025/7/

Matrix 1.15 released with OpenID Connect support, richer information about rooms before joining, better room discovery, additional text formatting options for messages, etc.:
alternativeto.net/news/2025/6/

7-Zip v25 released with faster compression, enhanced CPU support, improved legacy format handling:
alternativeto.net/news/2025/7/

HuggingChat is shutting down, will be replaced by another service better integrated into the Hugging Face ecosystem:
alternativeto.net/news/2025/7/

(more FOSS news in comment)

OdyseeIntroducing: Decentralized Payments on Odysee - For Everyone, EverywhereView on Odysee: Introducing: Decentralized Payments on Odysee - For Everyone, Everywhere

Cutting corners, reinventing the wheel, what could go wrong?!

I have not seen this in a long time.. It is modernity seemingly, they have to shut down the service in order to upgrade it.
Now, I am sure "AI/agentic" is NOT improving anything. Because it is Hugging Face, the face of "AI", they are at the forefront seemingly..

Hat hier jemand Erfahrung mit dem Betrieb eigener KI-Systeme? Nachdem der Chat von HuggingFace seine Tore sehr unerwartet und spontan geschlossen hat, bin ich auf der Suche nach freien Alternativen.
Eine Option könnte sein, so einen Chat mit den passenden LLMs selbst zu betreiben. Dafür wird dann aber eine vernünftige GPU nötig. GPU-Hosting scheint im Vergleich mit den gängigen V-Servern recht teuer zu sein. Zumindest für mich, ist das nichts, was ich ohne Finanzkonzept mal kurz buche ... Die günstigsten Preise lagen bei etwa um die 200 - 300 € pro Monat. Da scheint mir die Anschaffung eines lokalen Servers mit einer passenden Grafikkarte von Nvidia die mittelfristig günstigere Lösung.
Stimmt das? Übersehe ich dabei etwas?

Wie viele User können eigentlich auf so einem "simplen" Computer mit einer etwas besseren Grafikkarte parallel Anfragen stellen?

Hat hier einer / eine von euch Erfahrung und Tipps?

#HuggingFace #huggingchat #Ai #Ki @linux @linux #selfhost #linux #chat #bot #llm @askfedi_de #server