ervinxie

OpenAI plans to build its own AI chips on TSMC's forthcoming

Posted by ervinxie on Tuesday September 03, 2024 @09:07AM
from the pcgamer.com dept.

https://www.pcgamer.com/software/ai/openai-plans-to-build-its-own-ai-chips-on-tsmcs-forthcoming-16-nm-a16-process-node/

( Read More... | no comments | 3 points | on HN )

Re:Is My Blue Your Blue? on Monday September 02, 2024 @10:23PM (#41430661)

Is this test based on web rgb? I suggest use larger color space.

Re:Show HN: KTransformers–236B Model and 1M Context LLM Inference on L... on Thursday August 29, 2024 @09:05AM (#41390397)

The speedup is amazing!

Re:OpenAI is shockingly good at unminifying code on Thursday August 29, 2024 @07:20AM (#41389620)

LLMs are very good at text reading. LLMs read tokenized text, while human use eyes to view words. Another scenario is that ChatGPT is good at analyzing cpp template error messages, which are usually long and hard to understand for human.

Front End Strategies: Frameworks or Pure JavaScript?

Posted by ervinxie on Friday August 23, 2024 @07:48AM
from the thenewstack.io dept.

https://thenewstack.io/frontend-strategies-frameworks-or-pure-javascript/

( Read More... | no comments | on HN )

The Moss That Could Terraform Mars

Posted by ervinxie on Tuesday August 20, 2024 @07:22AM
from the nautil.us dept.

https://nautil.us/the-moss-that-could-terraform-mars-705890/

( Read More... | 1 comment | 2 points | on HN )

How to not die on (the way to) Mars

Posted by ervinxie on Friday August 09, 2024 @06:55AM
from the freethink.com dept.

https://www.freethink.com/space/mars-astronauts

( Read More... | no comments | 3 points | on HN )

Volunteers Complete a Simulated Year on Mars

Posted by ervinxie on Tuesday August 06, 2024 @07:52AM
from the universetoday.com dept.

https://www.universetoday.com/167778/volunteers-complete-a-simulated-year-on-mars/

( Read More... | no comments | on HN )

Seventh patient 'cured' of HIV: why scientists are excited

Posted by ervinxie on Thursday August 01, 2024 @07:09AM
from the nature.com dept.

https://www.nature.com/articles/d41586-024-02463-w

( Read More... | 1 comment | on HN )

Re:MoonshotAI unveils Kimi's large-scale LLM serving architecture on Tuesday July 02, 2024 @09:40AM (#40856552)

I have been wondering the reason why online generative AI can serving so many requests. This really gives me an explanation.