Apple is reportedly spending millions of dollars per day to train Ajax, its most advanced language model, which they believe to be more powerful than ChatGPT. Though Apple may seem late to the game on AI, don't underestimate a company with a $300B research war chest.
@aibreakfast
-
ChatGPT’s daily running cost estimated at $750k by credible sources
By
–
Revenue. Have heard from credible sources that the cost of running ChatGPT is around $750k/day.
-
ChatGPT Enterprise announces 32k token context for businesses
By
–
ChatGPT Enterprise just announced: 32k token context for businesses (!) https://
openai.com/blog/introduci
ng-chatgpt-enterprise
… -
The Tsunami of AI Will Completely Change the World
By
–
Knowing that the tsunami of AI will completely change the world.
-
Google’s Gemini to surpass GPT-4 with massive FLOPS increase
By
–
Gemini may eat the world: Google's Gemini is poised to surpass GPT-4 with a projected 5x more pre-training FLOPS by year's end and a potential 20x by next year. AI is just getting started.
-
AI algorithm converts thermal night vision to sharp color image
By
–
Found this super interesting: AI-aided imagery algorithm converts thermal night vision into sharp color image Conventional thermal vision (top) depicts this nighttime scene of a forest road in ghostly grays, and the system adds color based on the objects detected
-
LLMs training on self-generated text as machine learning inbreeding
By
–
LLMs generating new text that they will then use to further train LLMs is the machine learning version of inbreeding.
-
GPT-3.5 vs GPT-4 exam results at Wharton School
By
–
Graph of GPT-3.5 vs. GPT-4 results in various exams from Wharton School
-
Human-in-the-loop scalability via Mechanical Turk style intervention
By
–
Human-in-the-loop is more scalable than we might think. Mechanical Turk type intervention could be all you need. The models won’t be hung up on the entire edge case action, they may just need one piece of data to continue the decision tree.
-
LLMs run out of text, use AI to generate synthetic training data
By
–
LLMs are running out of text to train on So they are using the LLMs to create new training text