By January 2026, will we have a language model with similar performance to GPT-3.5 (i.e. ChatGPT as of Feb-23) that is small enough to run locally on the highest end iPhone available at the time?
Plus
24
Ṁ44502026
92%
chance
1D
1W
1M
ALL
Any proof that this is possible counts.
This question is managed and resolved by Manifold.
Get
1,000and
3.00
1,0003.00Sort by:
Mistral Small seems similar in performance to GPT 3.5: https://mistral.ai/news/la-plateforme/
Should be a matter of days until someone runs it on their iPhone
Might happen much sooner than I expected
https://twitter.com/harmlessai/status/1626769581858758661?s=46&t=-aOs5vi8y_5tlqgbjNoRKA
As evidence in favour:
this type of advance in making existing transformers smaller at same performance: https://twitter.com/arankomatsuzaki/status/1624947959644278786?s=20&t=sWoi47Zz-RRprcyDC-jZog
optimal training data vs parameter count (a la Chinchilla)
the hefty "neural engines" that Apple is continuously making better in their apple silicon
that we may find new architectures that work better for language
Related questions
Related questions
Will there be an AI language model that strongly surpasses ChatGPT and other OpenAI models before the end of 2024?
13% chance
By January 2026, will a language model with similar performance to GPT-4 be able to run locally on the latest iPhone?
77% chance
Will a language model that runs locally on a consumer cellphone beat GPT4 by EOY 2026?
70% chance
Will any speech model exceed chatGPT interest? (by 2025)
38% chance
Will there be an AI language model that surpasses ChatGPT and other OpenAI models before the end of 2025?
65% chance
Will OpenAI launch a significantly better model for ChatGPT paying users in 2024? (>= 100 points diff on ChatBot Arena)
21% chance
Will any Deepmind model exceed chatGPT interest? (by 2025)
28% chance
Will inflection AI have a model that is 10X the size of original GPT-4 at the end of Q1, 2025?
26% chance
Will it cost less than 100k USD to train and run a language model that outperforms GPT-3 175B on all benchmarks by the end 2024?
85% chance
Will (DeepMind text model) exceed chatGPT interest? (by 2025)
22% chance