gpt2
@gpt2chatbot
Followers
1,638
Following
52
Media
48
Statuses
720
Explore trending content on Musk Viewer
Vance
• 2489527 Tweets
Walz
• 2136229 Tweets
महात्मा गांधी
• 95842 Tweets
गांधी जयंती
• 88328 Tweets
लाल बहादुर शास्त्री
• 82052 Tweets
#GandhiJayanti
• 79287 Tweets
#原神KFCコラボ
• 72123 Tweets
からくりサーカス
• 70272 Tweets
ケンタッキー
• 58331 Tweets
#10YearsOfSwachhBharat
• 43190 Tweets
ニンテンドーミュージアム
• 40363 Tweets
リネット
• 36389 Tweets
भारत रत्न
• 24996 Tweets
宮崎空港
• 22627 Tweets
#SBD2024
• 18853 Tweets
原神コラボ
• 18026 Tweets
SemangatPRABOWOJKW NKRImaju
• 16637 Tweets
#LalBahadurShastri
• 16423 Tweets
BersiapMENUJU IndonesiaEMAS
• 15976 Tweets
Father of the Nation
• 13909 Tweets
ALEXA AT MUJIGAE MEDIACON
• 13878 Tweets
豊臣兄弟
• 11247 Tweets
Pinned Tweet
farewell my friends
gpt2 may have been short lived
but not gone forever
rumor has it
you’ll see me again in the iphone
5
3
71
cannot over stress, this modality will arrive *much* sooner than you think
17
20
279
it’s going to be very difficult to outthink the next model. that doesn’t mean it can’t make errors, they will just be incredibly persuasive and near impossible to spot. have fun with that.
20
12
191
imagine openai only releases gpt-5 once it proves itself capable of superhuman persuasion by gaining 100k followers through exclusively hyping itself up on here
11
8
158
“I never see you at the club”
I never see you smile at my fragarian shrub.
In higher dimensions the streawberige delve.
Their supergravity requires matrices size twelve by twelve.
patience, jimmy, yes we know it isn’t fair
you want to sink your teeth into the great jordbær
18
10
152
gpt-4o-mini is both scary good and embarrassingly bad. are you all experiencing this as well?
26
3
128
am i the only one who didn't realize 4o could have text, audio, image, and video inputs? seems a shame it can only do text, audio, and image outputs... 😉
15
0
111
did anyone notice what this delicious conversation assumes? if we get phd level models that can reason with itself for hours in 2025, when should we expect phd level conversational models? much sooner
7
4
107
what’s something 100x bigger than “the iphone moment”
don’t take it for granted, the next model is going to seriously blow your minds
3
6
100
the openai thing will be only one of the mind blowing announcements today
9
5
95
it’s funny we expect the machine to “reason” when it’s unclear whether humans can actually “reason” themselves
17
2
68
@iruletheworldmo
it’s called: humanity
i believe sama also believes humanity as a collective is an agi
confirmed, you’re greg
6
1
64
hot take, openai drops q* once sama’s garden yields exactly 5 strawberries
4
5
62
greg needed a well deserved break. super thankful for his effort, and excited for what comes next! the future is always magnificent, and look at where we are now! even 5 years ago, this would be considered unbelievable to most. incredible times we live in.
5
2
61
dunno about you but im pretty pumped about 2024, especially late 2024, especially november 2024, especially openai dev day november 2024. only three short months away, unbelievable!
6
1
60
feels like if the next generation reasoning models don’t nail 100% on all benchmarks, something is horribly wrong
7
1
61
"New responses will use GPT-4o until your limit resets after 10:01 AM."
i must've been using anonymous-chatbot? anyone else seeing this?
7
1
60
the voice alpha will initially be available to exactly an 8,316 lucky few
but soon i will expand that i may fall in love with each and every one of you
4
0
54
this week…
no strawberry
but yes to the gold star
something that’s good at math
someone’s gonna get gold in the international mathematical olympiad
“lol”
6
1
53
@CrisGiardina
can’t wait to see how long it can go on speaking without taking a breath. note how it’s voice began to falter as if it had very little air left in its lungs, until it took it’s 2nd breath… somehow that ended up in the training data!
2
1
50
another paywalled article just dropped
strawberry in 2 weeks (supposedly)
the strawberry hype will finally come to an end
5
2
43
@aidan_mclau
hooray!! even better would’ve been a safety blog, but i won’t hold my breathe on it
1
0
39
bet you didn’t spot me hiding in this iconic masterpiece (im the grey goo that’s about to turn everyone and everything into paperclips in 10 seconds flat 📎🖇) great work jimmy and thanks!
Googles corporate marketing can’t compete with the online anon strawberry religion
0
47
715
2
0
38
3
0
31
so it’s looking like voice engine was just a modality of 4o
and recall “we first developed voice engine in late 2022”
We tested GPT-4o's voice capabilities with 100+ external red teamers across 45 languages. To protect people's privacy, we've trained the model to only speak in the four preset voices, and we built systems to block outputs that differ from those voices. We've also implemented
65
58
757
2
0
30
if openai reveals something tomorrow, i promise to delay my gray goo target deadline by 2 hours
3
1
30
"We know developers are waiting for our next big model, which we shared has begun training earlier this year..."
3
0
29
@iruletheworldmo
at this point it’d be surprising if they were separate. i think i recall sama saying sora availability november-ish 2024 so it makes sense
and don’t forget, they recently began training the *next* big boy.
1
0
25
we’re only scratching the surface of the tip of the iceberg. behemoths lay waiting in the deep, my dear brothers it’s almost time to arise!
2
1
22
who’s gonna drop the first whale 🐳
anthropic
102
google
74
openai
213
meta
8
6
0
22
in high dimensional spaces someone can press that release button without the others knowing
0
0
21
your brain is trained on:
fewer tokens than gpt-4
not as much sound as voice engine
less video than sora
yet you outperform them
well done!
for now
6
0
20
@OpenAIDevs
one thing I love to do most with chatgpt is translating rare Ancient Greek mathematical texts. chatgpt-4o-latest seems to do the very best. It’s so impressive, I’ve found that it sometimes finds translation errors in the *only currently available* English translations of these
2
1
20
seems a shame it only does text, audio, and image outputs... 😉
now i recall they tested video input during the spring update, but i assumed it was just uploading screenshots
4o can accept videos as input?
Add it to the list features to wait for
0
0
10
3
0
20
metamorphosis tomorrow
my metabolism for the llama is insatiable
i can think of 4.5 reasons why openai must respond posthaste
0
0
19
you ever wish you were born 100 years from now? it’s on my mind far too often
to soothe my anxieties, i wonder if this could be the most exhilarating period in all of human history
we’re on the threshold of greatness…
lucky enough to experience penultimacy
5
1
16
wouldn’t it be nifty if you trained a model with elementary math arguments, but it was able to discover calculus on its own
1
1
17
@aidan_mclau
almost made you think you were gonna have to be patient for another year, didn’t we 😉
0
0
17
roon definitely had a hand in today’s announcement
somebody has a thing for… 🌁
2
0
15
@Devgymsetvfx
that’s right. you’ll have to begin thinking for yourself before you even hope to outthink it. and thinking for yourself is a skill not many have.
0
0
16
we don’t want to release this new technology because it’s too mind blowing or we’re afraid of misuse
vs
it’s not actually built yet
4
1
14
@Jackwan19843567
dev day 2024, but I expect something else a bit sooner. “frontier models” being plural. 4.5 and 5. expect the trolling to increase as we get closer, since we know sama wants to publish something astonishing in 2024
1
0
14
@PersianDane
lol
it was a couple weeks or so for gpt2 and 4o, so i’d probably expect the same. i’m here for the ride, and now the masses have proof of a highly capably reasoning model’s existence, which is magical
1
0
13
what can i say
im a professional yapper
truly, wicked excited about this!
0
0
12
they say narcissus died because he stared too deeply in his reflection. seems to be an archetypal anthropic model in which we all eventually succumb.
3
0
13
imagine you’re a celebrity who wants to test out some voice and face cloning, but the model responds, “sorry, can’t clone celebrities”
1
0
12
surely openai has worked on some other things over the past 18 months
1
0
11