Recent searches

No recent searches

Search options

Only available when logged in.
101010.pl is one of the many independent Mastodon servers you can use to participate in the fediverse.
101010.pl czyli najstarszy polski serwer Mastodon. Posiadamy wpisy do 2048 znaków.

Administered by:

Server stats:

504
active users

101010.pl: AboutStatusProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.4.0-beta.2

#llama

2 posts2 participants1 post today
Replied in thread
*
c_th1

@netzpolitik_feed

Ich habe euren Post mal etwas Überarbeitet, damit er spannender klingt.....

,, Der #META Konzern will die Daten von #Facebook, #Instagram, #WhatsApp und CO zum Training seiner #KI - #MetaAI - #Llama, benutzen.

Einspruch noch bis zum 26.05.25 möglich, sonst werden deine Daten fürs KI Training genutzt.
Oder noch besser, alle diese Dienste nicht mehr benutzen ,,

netzpolitik.org/2025/meta-ki-j

@Bundesverband

netzpolitik.org · Meta-KI: Jetzt widersprechen oder für immer schweigenMark Zuckerberg will wieder mal an unsere Daten - dieses Mal um seine KI zu trainieren. Bis zum 26. Mai ist Widerspruch möglich: Wir erklären, warum das eine gute Idee ist, und zeigen, wie es funktioniert.
Continued thread
*
Nonilex

Update:

Judge in #Meta case warns #AI could 'obliterate' market for original works

A skeptical federal judge in SF on Thurs questioned Meta Platforms' argument that it can legally use copyrighted works without permission to train its #ArtificialIntelligence models.
In the 1st court hearing on a key question for the AI industry, Judge Vince Chhabria grilled attys for both sides over Meta's request for a ruling that it made #FairUse of books to train its #LLaMa #LLM.
#law
reuters.com/legal/litigation/j

Continued thread
Nonilex

#Meta told Chhabria in March that it used the #authors' material “transformatively”, to teach #LLaMa to "serve as a personal tutor on nearly any subject, assist with creative ideation, & help users to generate business reports, translate conversations, analyze data, write code, & compose poems or letters to friends."

"What it does not do is replicate Plaintiffs' books or substitute for reading them," Meta said.

#law#tech#AI
Continued thread
*
Nonilex

The #FairUse question hangs over lawsuits brought by #authors, #news outlets & other #copyright owners against companies including #Meta, #OpenAI & #Anthropic. The #legal doctrine allows for the use of copyrighted work without the copyright owner's permission under some circumstances.

The authors in the Meta case sued in 2023, arguing the company used pirated versions of their books to train #LLaMa without permission or compensation.

#law#tech#AI
*
Nonilex

Judge in #Meta case weighs key question for #AI #copyright lawsuits

A federal judge in SF will hear arguments on Thurs from Meta & a group of #authors in the 1st court hearing over a pivotal question in high-stakes copyright cases over #AItraining.

Judge Vince Chhabria will consider Meta's request for a pretrial ruling that it made "#FairUse" of books from writers including Junot Diaz & comedian Sarah Silverman to train its #LLaMa #LLM.

#law #tech #IntellectualProperty
reuters.com/legal/litigation/j

st1nger :unverified: 🏴‍☠️ :linux: :freebsd:

Proof-of-concept project, showing that it's possible to run an entire Large Language Model in nothing but a #PDF file.

It uses #Emscripten to compile #llama.cpp into asm.js, which can then be run in the PDF using an old PDF JS injection.

Combined with embedding the entire #LLM file into the PDF with base64, we are able to run LLM inference in nothing but a PDF

github.com/EvanZhouDev/llm.pdf

GitHubGitHub - EvanZhouDev/llm.pdf: Run LLMs inside a PDF file.Run LLMs inside a PDF file. Contribute to EvanZhouDev/llm.pdf development by creating an account on GitHub.
TrendingLive feeds
About