Lifeboat Foundation

Information and reality metaphysics and logic.

Shared with Dropbox.

The philosopher’s job is surely more than to be a servant to the sciences

Meta’s move into the open social web, also known as the fediverse, is puzzling. Does the Facebook owner see open protocols as the future? Will it embrace the fediverse only to shut it down, shifting people back to its proprietary platforms and decimating startups building in the space? Will it bring its advertising empire to the fediverse, where today clients like Mastodon and others remain ad-free?

One possible answer can be teased out of a conversation between two Meta employees working on Threads and Flipboard CEO Mike McCue, whose company joined the fediverse with its support of ActivityPub, the protocol that powers Mastodon and others.

On McCue’s “Flipboard Dot Social” podcast, he spoke to two leaders building the Threads experience, director of product management Rachel Lambert and software engineer Peter Cottle. McCue raised questions and concerns shared by others working on fediverse projects, including what Meta’s involvement means for this space, and whether Meta would eventually abandon Threads and the fediverse, leaving a destroyed ecosystem in its wake.

How AI powered chatbots affect human conversations and the dominance of chat for Generation Z. The future of communications.

Tariffs and IRA tax incentives are starting to reshape global supply chains—but vast challenges lie ahead, explains Shawn Qu, founder of Canadian Solar.

Tiny crystals grown within molten metals at atmospheric pressure.

The team discovered the alloy’s surprising properties and then figured out how they arise from interactions in the atomic structure.

As per the team of researchers, the alloy is from a new class of metals known as refractory high or medium entropy alloys (RHEAs/RMEAs).

Meta presents Layer Skip.

Enabling early exit inference and self-speculative decoding.

We present LayerSkip, an end-to-end solution to speed-up inference of large language models (LLMs).

Continue reading “Layer Skip: Enabling Early Exit Inference and Self-Speculative Decoding” »

Meta presents MoDE

CLIP Data Experts via Clustering.

The success of contrastive language-image pretraining (CLIP) relies on the supervision from the pairing between images and captions, which tends to be noisy in web-crawled data.

Continue reading “MoDE: CLIP Data Experts via Clustering” »