touringa
Re:Elon: Code is brittle. Need a complete rewrite
on Monday March 06, 2023 @06:36PM (#35049435)
Re:Eventually ChatGPT will eat itself
on Wednesday February 22, 2023 @12:08AM (#34891890)
It's already happening.
“ChatGPT, a version of OpenAI’s GPT-3.5 model… gained more than 100m users in its first two months, and is now estimated to produce a volume of text every 14 days that is equivalent to all the printed works of humanity.”
— Dr Thompson, Feb/2023, cited in report by the National Bureau of Economic Research (Scholes, Bernanke, MIT)
https://www.nber.org/system/files/working_papers/w30957/w309...
Re:Outage on All Models
on Tuesday February 21, 2023 @05:35AM (#34879335)
Just for clarity, this is Azure[1].
[1] https://venturebeat.com/ai/openai-is-reducing-the-price-of-t...
Re:Ask HN: Does Google use the text inside gdocs and Gmail for trainin...
on Wednesday February 15, 2023 @03:28PM (#34810091)
If you're talking about Google Bard, they were very clear in the LaMDA 2 paper that they only used public sources.
"...from public dialog data and other public web documents..."
LaMDA 2 paper: https://arxiv.org/abs/2201.08239
My overview of Google Bard including dataset: https://lifearchitect.ai/bard/
My overview of Google PaLM and Pathways family including dataset: https://lifearchitect.ai/pathways/
Compare with other models including the use of DeepMind's MassiveWeb/MassiveText and EleutherAI's Pile dataset: https://lifearchitect.ai/whats-in-my-ai/
Re:Ask HN: Does Google use the text inside gdocs and Gmail for trainin...
on Wednesday February 15, 2023 @03:24PM (#34810036)
Actually, it's 'BookCorpus'. OpenAI spelt it wrong in their GPT-1 paper.
It has also been analyzed here and here:
Re:Dijkstra comparing computer science to alchemy (1989)
on Sunday February 12, 2023 @10:14PM (#34769744)
Ilya used that term a lot, and continues to do so.
https://www.newyorker.com/magazine/2019/10/14/can-a-machine-...
https://twitter.com/ilyasut/status/1486728489319362571?lang=...
Re:The entire prompt of Microsoft Bing Chat
on Thursday February 09, 2023 @04:33PM (#34731839)
Full text:
https://lifearchitect.ai/bing-chat/
Re:
on Sunday February 05, 2023 @07:15AM (#34664283)
ChatGPT is probably outputting at least 110x the equivalent volume of Tweets by human Twitter users every day.... and the equivalent of the entire printed works of humanity every 14 days (Jan/2023)
Re:Ask HN: What were the papers on the list Ilya Sutskever gave John C...
on Friday February 03, 2023 @12:45PM (#34644830)
Re:ChatGPT Cheat Sheet
on Tuesday January 24, 2023 @11:53AM (#34506258)
The ChatGPT Prompt Book:
https://docs.google.com/presentation/d/17b_ocq-GL5lhV_bYSShz...
Re:ChatGPT killer: Google-owned, UK-born AI startup Deepmind is workin...
on Monday January 23, 2023 @01:53PM (#34493221)
Re:Meet “Claude”: Anthropic’s Rival to ChatGPT
on Wednesday January 18, 2023 @08:19AM (#34426274)
lol!
Excellent candidate for https://www.reddit.com/r/dontyouknowwhoiam/
Re:Meet “Claude”: Anthropic’s Rival to ChatGPT
on Wednesday January 18, 2023 @08:17AM (#34426260)
Video demo: https://youtu.be/B7Mg8Hbcc0w
More info on Claude's principles/Constitution: https://lifearchitect.ai/anthropic/
Re:GPT-3.5 and Wolfram Alpha via LangChain by James Weaver
on Tuesday January 17, 2023 @10:00PM (#34422930)
Video walkthrough and interview with creator James Weaver: https://youtu.be/wYGbY811oMo
Re:Ask HN: How does ChatGPT work?
on Sunday December 11, 2022 @02:07AM (#33940878)
Re:
on Saturday December 10, 2022 @09:16PM (#33939361)
The human feedback fine-tuning concept was applied following strict policies and rules. The rules chosen by OpenAI would be very similar to those applied by DeepMind for the Sparrow dialogue model (Sep/2022), which is a fine-tuned version of DeepMind’s Chinchilla model.
The rules used for DeepMind Sparrow were selected by researchers from DeepMind (Alphabet), California Institute of Technology, University of Toronto, and University College Dublin.
Re:DeepMind Dramatron: a new tool for writers to co-write theatre and ...
on Friday December 09, 2022 @04:09PM (#33926482)
This is an older paper, and demo'd here with GPT-3 instead of Chinchilla.
Video of DeepMind Dramatron (Chinchilla version):
Re:Ask HN: When/where can we get an offline GPT3 type chatbot?
on Wednesday December 07, 2022 @11:13PM (#33904058)
Here's a full list of the big ones:
https://lifearchitect.ai/models/
https://lifearchitect.ai/timeline/
There are another 97,000 Transformer models on HF:
Re:Yandex YaLM 100B: Largest GPT-Like Neural Network in Open Source
on Sunday December 04, 2022 @06:29PM (#33859239)
Maybe back in June 2022.
Since then, around 70,000 new Transformer models have been released via HF.
There are several models much larger than YaLM, including BigScience BLOOM 176B and Tsinghua GLM-130B.
Re:I made ChatGPT take an SAT Test
on Saturday December 03, 2022 @01:50AM (#33841320)
In 2020, GPT-3 175B was already outperforming humans on SAT questions by ~20%. There are far more formal benchmarks for models now.
Re:Twitter’s $5B-a-year business hit as Elon Musk clashes with adverti...
on Sunday November 27, 2022 @02:42AM (#33761215)
Re:Galactica: an AI trained on humanity's scientific knowledge (by Meta)
on Wednesday November 16, 2022 @12:25AM (#33619264)
Re:
on Wednesday November 02, 2022 @03:43AM (#33432423)
A list of GPT-3 and Codex clients was announced by Microsoft’s Satya Nadella and OpenAI’s Sam Altman in October 2022:
Accenture, AI Busters, Alert Innovation, American Fidelity, Aon, Arvato Bertelsmann, Autodesk, Avanade, BMW, Carmax, Cipio.ai, Clevertar, Cognizant, Databook, Downer, EY, Farmlands Cooperative, Genie AI, Elsevier Health, HSBC, IFAD, Inpris, Inward, Kept, Klaviyo, Morgan Stanley, National Nederlanden, PwC, Shell, Snelstart, Sogeti, Soul Machines, Strabag, Trelent, Wipro, WordLift, Zammo.ai
Re:Ernie-ViLG better anime quality than Stable Diffusion
on Friday September 02, 2022 @02:39AM (#32687332)
Very high-level...
Models: https://lifearchitect.ai/models/
Timeline: https://lifearchitect.ai/timeline/
Text-to-image: https://lifearchitect.ai/art/
Re:Transformer Models: An Introduction and Catalog
on Friday July 22, 2022 @07:12AM (#32190705)
Re:CogVideo: Large-Scale Pretraining for Text-to-Video Generation via ...
on Monday May 30, 2022 @03:31PM (#31562726)
Re:Imagen, a text-to-image diffusion model
on Monday May 23, 2022 @06:32PM (#31485669)
SymphonyNet: https://youtu.be/m4tT5fx_ih8
Re:A Man Who Controls Computers with His Mind
on Saturday May 14, 2022 @08:02AM (#31377138)
Re:Sir David Attenborough's Address to World Leaders at COP26 [video]
on Thursday November 04, 2021 @03:44PM (#29111673)
Dude is 95 years old, and one of the most effective public speakers I have ever seen.