Hacker Newsnew | past | comments | ask | show | jobs | submit | more rayval's commentslogin

I have used Microsoft Office for 20 years and Google Docs/Workspace for 15 years. I much prefer the Google suite over the Microsoft one. I prioritize quality of user experience, ease of use, performance, platform portability, freedom from constant intrusive software updates, and not messing with the built-in Emacs key bindings built into every Mac (Various MS Office programs inconsistently disable those key bindings). YMMV, esp with UI and UX, but for me there's no contest.

Regarding support, that is not an issue I have never needed it for Google, but occasionally have needed for MS Office. So I can see why if you are using Office you would value that.


To be clear, this is just a rumor passed along by Bindu Reddy. who I consider pretty well informed. Anyone have more detailed sources?


She's quoting from https://www.theinformation.com/articles/meet-mai-1-microsoft... without citing it for some reason.


Here's a compelling visualization of the functioning of an LLM when processing a simple request: https://bbycroft.net/llm

This complements the detailed description provided by 3blue1brown


When visualised this way, the scale of GPT-3 is insane. I can't imagine what 4 would like here.


IIRC, GPT-4 would actually be a bit _smaller_ to visualize than GPT3. Details are not public, but from the leaks GPT-4 (at least, some by-now old version of it) was a mixture of expert, with every model having around 110B parameters [1]. So, while the total number of parameters is bigger than GPT-3 (1800B vs. 175B), it is "just" 16 copies of a smaller (110B) parameters model. So if you wanted to visualize it in any meaningful way, the plot wouldn't grow bigger - or it would, if you included all different experts, but they are just copies of the same architecture with different parameters, which is not all that useful for visualization purposes.

[1] https://medium.com/@daniellefranca96/gpt4-all-details-leaked...


Mixture of Experts is not just 16 copies of a network, it's a single network where for the feed forward layers the tokens are routed to different experts, but the attention layers are still shared. Also there are interesting choices around how the routing works and I believe the exact details of what OpenAI is doing are not public. In fact I believe someone making a visualization of that would dispell a ton of myths around what are MoEs and how they work


amazing visualisation


Intriguing comment that I don't quite follow. Can you please elaborate?


Probably OpenAI running on Azure. But it was still convoluted.


Also I have seen some papers on Arxiv from FAANG companies that have a half-dozen or more co-authors, and almost all of those authors have Chinese names.


This is flimsy propaganda with a ham-handed political agenda. This video is not alone. Multiple "observers" have been predicting the imminent collapse of China, starting 25 years ago.

Here's a list of 25 such predictions in an almost yearly cadence: https://x.com/rippinkitten18/status/1701411830642741348?s=20

In the meantime, China has outpaced most Western countries with steady growth.

One notable "expert" is Paul Krugman, who predicted Chinese doom a couple of months ago. Turns out he made a similar prediction ten years ago: https://x.com/NuryVittachi/status/1701400962957746544?s=20

For an accurate street view about Beijing recorded over the past 2 weeks, here are two random videos that show a bustling and prosperous Beijing, taken by Western travelers:

https://www.youtube.com/watch?v=hFPAdqLKVRY

https://www.youtube.com/watch?v=eC2sGM7CqmM


This one is from Falun Gong which is in a fascinating category of its own.



Google Bard also gave a clear answer: https://imgur.com/a/8HGivir


They also censored critics of Modhi's government in India.


On Twitter, I never use to get spam DMs but now it is a daily occurrence. I never blocked anyone over 10 years but now I am blocking multiple people per day. I used to never see anti-semitic content, but now I run across it regularly (even though I don't follow similar accounts). I still enjoy interesting discussions on crypto and genAI tech/trends but that is probably not a broad enough audience to be sustainable for Twitter/X.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: