Artificial Intelligence+anon exchanges 17
.
Interested to know where people stand on this. Should technological progress be pursued regardless of any regulation?: https://bensbites.beehiiv.com/p/inside-eacc-new-religion-silicon-valley?utm_source=bens[…]etter&utm_campaign=daily-digest-maybe-it-s-all-about-moneyhttps://www.theinformation.com/articles/its-a-cult-inside-effective-accelerationism-the-[…]_campaign=inside-e-acc-the-new-religion-in-silicon-valley
Ben's BitesSome think it’s the purpose of their life (and all humans), few think it’s a passing phase while others feel it’s more of a fanatic cult. (854 kB)
The Information“Pharma Bro” Martin Shkreli, Y Combinator president Garry Tan and Notion co-founder Chris Prucha looked up at the cartoon of a shirtless man, his six-pack abs gleaming against a swirling galactic background. They were eager to listen to what the man had to say. The cartoon was the avatar for ...
Well, I am for the freedom of research which is an important right since 'NGOs' even try to make technologies 'bad' with complete lies (Like Greenpeace with GMO).The complete AGI warnings are basically as 'science based' like flying atomic car predictions in the 70ties. It is just following self interests to have more 'control' but looking how 'correct' Werner von Braun was as Nasa Boss with flying cities in space, it seems completely overblown with this AGI discussion. This is like people thinking CERN if producing a black hole and destroying earth.https://www.bbc.com/news/technology-65886125 Meta AI lead about this... This is one of the three godfathers of AI after all.In the end, the system cannot have 'unlimited revolution anyway since it otherwise decays in 'mass war' clearly before you reach 'singularity'.
Tuesday, October 10th
https://www.learn.xyz/ <- fun little product, using AI
Learn.xyz: Learn anything, in any language
Use the power of AI to create lessons, share them, and inspire others. How long will your streak be? (215 kB)
Wednesday, October 11th
MediumThe definitive guide for choosing the right method for your use case
Reading time
19 min read
Aug 29th (124 kB)
X (formerly Twitter)Zephyr alpha, a Mistral fine-tune (using DPO and Ultrachat), achieves results similar to Chat Llama 70B in multiple benchmarks (ARC, HellaSwag, MMLU, TruthfulQA) and above results in MT Bench
Quite exciting to see 7B models with such high performance!https://t.co/bTtHveyT2F (32 kB)
Quite exciting to see 7B models with such high performance!https://t.co/bTtHveyT2F (32 kB)Thursday, October 12th
https://finance.yahoo.com/news/amd-expands-open-ai-software-155000826.html?guccounter[…]tAJOwoFi_u2hkSNpp2ja_oqek0KasbdZPCaAFMGpaK4KHCQVhr5KczFMtrZlhttps://gizmodo.com/github-copilot-ai-microsoft-openai-chatgpt-1850915549Future AI, business models and costs... It will be interesting... but new driver software may be the way since we not know how good 'ai' chips still can get with Moore's law. (1nm or similar is the end...) Problem is that GPU got much faster in the last years than Moore's law predict but not if you look on the price for each calculation...
<img src="https://ca.slack-edge.com/T02E1FAT8-U043V2L7VPW-24d7cb830caf-48" alt=""" aria-hidden="true" /">
Andrei Popov 4:35 AM
Reading about Sepp Hochreiter’s xLSTM…probably it is just not the good time to invest into any LLMs for the next year. I mean, it’s like pip trading right now - many changes inside of the day you can make use of, but a lot of energy is needed and losing some “transactions” is inevitable. Mid term it feels to stay put and observe the market is much better
(edited)
(edited)
<img src="https://ca.slack-edge.com/T02E1FAT8-U043L0S217Y-f5047cbd6d01-24" alt=""" aria-hidden="true" /"><img src="https://ca.slack-edge.com/T02E1FAT8-U043V2L7VPW-24d7cb830caf-24" alt=""" aria-hidden="true" /"> 8 replies
Last reply 6 days agoView thread
Friday, October 13th
This is interesting, Meta has paid Kendall Jenner to create an AI chatbot. @yoursisbillie
IMG_0196
Lightning AILoRA is one of the most widely used, parameter-efficient finetuning techniques for training custom LLMs. From saving memory with QLoRA to selecting the optimal LoRA settings, this article provides practical insights for those interested in applying it.
Written by
Sebastian Raschka
Est. reading time
15 minutes
Oct 12th
Monday, October 16th
Some fantastic stuff coming from Canva!https://www.canva.com/magic/
.