Could DeepSeek R1-Lite-Preview, o1-Preview, Claude Sonnet 3.5, or Gemini 1121 Work at Google? (LIVE)

[@jeffwads]

These questions exposed the mathematical abilities of these models quite well. That Deepseek R1 is impressive.

[@KyleKabasares_PhD]

Thanks so much! DeepSeek has really impressed me!

[@martini1179]

It's good to see you back. I love how you're carving out a niche using your PhD and mathematical skills in order to test these models. Also, your fun episodes are also fun, like the ones with your grandmother, or when you were messing around with advanced voice mode. Most people just make unimaginative requests. I wish you all the best.

[@KyleKabasares_PhD]

Thank you for your kind words!

[@BilalAzhar-f2t]

Chat this guy is who wholesome i cant even troll properly

[@parthasarathyvenkatadri]

If you change the numbers in the dedicated prompt does it still get it ...

[@bobsalita3417]

If the LLM arrives at the correct answer, ask again but substitute airplanes for trucks.

[@riothecat]

Kyle, I asked DeepSeek if it knew who you are. Here’s the first sentence: Kyle Kabasares is an American actor known for his role as Private First Class Greg "Pig-Pen" Colburn in the television series "Tour of Duty" (1987-1990). [That’s a hallucination]

[@johnnyBrwn]

Could be another Kyle. If you ask someone if they know "ice tea" they might mistake you talking about beverage or the rapper.

[@eternalroree]

love these streams

[@darwindoppelganger8224]

The GPT models remember your chats from one session to another, perhaps biasing your test results. Ask GPT about "turning off Memory".

[@xinpingdonohoe3978]

Does it? That's...something. If you turn it off, does it lobotomise it and remove its memory of what you said, or just stop it accumulating any more data?

[@nashh600]

That's not exactly true, it can chose what it remembers but it's never entire conversations and from my experience it's very rare that it puts something in its memory unless you explicitly ask

[@procrastinatingrn3936]

Im a semster 1 cs student and need help at maths would any of these be a good tutor , i want to learn not to solve the problems

[@parthasarathyvenkatadri]

Can i calculate The 50 truck question ... as lets say we travel 50 miles then half of all the trucks fuel will be empty ... Now if we shuffle the half fuel from half the trucks to fill the other half .. we would have 25 fully fuelled trucks ... And then after travelling another 50 miles we 12.5 fully fuelled trucks ... Then tge next 50 will have 6.25 then next 50 will have 3.125 , the next 50 will be 1.5625 ..so we could then say we can travel 156 miles more so 406 ... Apporximately ... What am i missing ... how am i off by about 50 miles

[@parthasarathyvenkatadri]

Can it be solved like this ... Lets connect all the trucks together .... So if there are 10 trucks and each truck independently can travel 100 miles on a full tank ... If we connect all the trucks and run them using just the first truck until it is empty it would have 1/9 of its efficiency ...then we disconnect one truck and run again now with the next truck going through its fuel ...

[@homeyworkey]

Entire book of questions to ask and ur just reformatting one problem for an hour thats wild. I get it being in the training, but for an entire hour, just reformat it once

[@BilalAzhar-f2t]

Make a discord server please if that is something you would like to do

[@KyleKabasares_PhD]

@@BilalAzhar-f2t I already have!

[@blinkers88]

he already has one

[@KyleKabasares_PhD]

You can join here: discord.gg/YpzdbKsC

[@wyqtor]

Damn, DeepSeek totally pwned OpenAI, even with a limited compute budget due to sanctions 😮

[@blinkers88]

pwned lol I haven't seen that word in 20 years

[@wyqtor]

@@blinkers88 It's coming back, Pliny uses it all the time in his jailbreaks.

[@haiderameer9473]

It really makes you wonder how they would perform if they weren’t compute constrained

[@EssentiallyAI]

The sanctions haven't had an impact yet.

[@madushandissanayake96]

Am i the only one notice that these questions are dumb?

[@mirek190]

yes only you

[@madushandissanayake96]

@mirek190 I think I am a genius then