trieu
@thtrieu_
Followers
2,529
Following
135
Media
11
Statuses
1,200
inventor of #alphageometry . lead of alphageometry 2. thinking about thinking @ deepmind.
Mountain View
Joined April 2014
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
#TGS2024
• 192887 Tweets
Miami
• 102699 Tweets
Knicks
• 86300 Tweets
Towns
• 74987 Tweets
Minnesota
• 71529 Tweets
Randle
• 66334 Tweets
León
• 66257 Tweets
mingyu
• 53514 Tweets
भगत सिंह
• 50906 Tweets
Donte
• 41552 Tweets
Wolves
• 39800 Tweets
Paola
• 31972 Tweets
The ACC
• 30055 Tweets
DiVincenzo
• 29603 Tweets
#DAY6INBKK2024
• 28937 Tweets
日本破壊クソメガネ
• 22114 Tweets
Virginia Tech
• 20101 Tweets
Cam Ward
• 19536 Tweets
#キントレ
• 17221 Tweets
Canes
• 15528 Tweets
महान क्रांतिकारी
• 14382 Tweets
Rutgers
• 13347 Tweets
ババコンガ
• 11639 Tweets
#bhagatsinghjayanti
• 10825 Tweets
#AFLGF
• 10023 Tweets
Pinned Tweet
happy to contribute!
Powered with a novel search algorithm, AlphaGeometry 2 can now solve 83% of all historical problems from the past 25 years - compared to the 53% rate by its predecessor.
It solved this year’s IMO Problem 4 within 19 seconds. 🚀
Here’s an illustration showing its solution ↓
6
38
371
5
5
120
Proud of this work. Here's my 22min video explanation of the paper:
Introducing AlphaGeometry: an AI system that solves Olympiad geometry problems at a level approaching a human gold-medalist. 📐
It was trained solely on synthetic data and marks a breakthrough for AI in mathematical reasoning. 🧵
126
1K
4K
39
154
770
We show large language models trained on massive text corpora (LM1b, CommonCrawl, Gutenberg) can be used for commonsense reasoning and obtain SOTA on Winograd Schema Challenge. Paper at , results reproducible at
7
92
303
As also observed by OpenAI's GPT-2, training data quality is important. We release the STORIES corpus introduced in our work . The corpus is a high quality subset of CommonCrawl with a total of ~7B words (~32GB) can be found here:
3
47
178
Wow! An old project of mine is now the 7th most popular Machine Learning project across all Github in 2018, alongside with Tensorflow and Scikit-learn? I really need to spend some time polishing it now...
From the programming languages you used most to the most popular data science packages, we’re digging into the data on Machine Learning from 2018. Find out what we discovered
0
76
213
3
1
64
Human reasoning is not manipulating symbolic expressions
5
17
42
Had the chance to sit next to Daniel
@xpearhead
in the early days of the project and tried out the interactive Meena. It has always been *this* surprising and funny :) BIG Congrats to the team with this publication. The possibilities to build up from here is endless.
1
4
33
Excited to share a new work by
#GoogleAI
resident
@thtrieu_
(with
@iamandrewdai
, me, & Quoc Le) on training very long RNNs (up to 16K long). See paper for extreme cases of zero or little backprop on RNNs ;)
0
45
163
0
9
30
"We see contributions to traditional conferences and publications in journals as an important part of our work, but also support efforts that go “beyond the research paper"".
I'm excited to finally share what I have been working on.
Today we are officially launching Cohere For AI
@forai_ml
a non-profit research lab that aims to reimagine how, where, and by whom research is done.
63
118
820
0
1
13
A commonsense reasoning task is "solved" even before its official introduction.
It's amazing how fast
#NLProc
is moving these days.
We have now reached super-human performance on SWAG, a commonsense task that will only be introduced at
@emnlp2018
in November!
We need even more challenging tasks!
BERT:
SWAG:
8
87
294
0
1
11
10M toks with almost perfect needle in haystack, amazing.
Gemini 1.5 Pro - A highly capable multimodal model with a 10M token context length
Today we are releasing the first demonstrations of the capabilities of the Gemini 1.5 series, with the Gemini 1.5 Pro model. One of the key differentiators of this model is its incredibly long
195
1K
6K
1
0
7
Yes, the fact that we can do this means the geometry we considered is quite narrow.
Making a synthetic dataset of mathematical proofs is hard! It's easy to make a whole lot of "1+1+1+...=491" style theorems.
I'm surprised this method of random construction and transformation finds so many classical geometric theorems.
Maybe because the domain is somewhat
4
5
50
0
0
7
Want to see space-time contract/dilation? This series on Special Relativity (SR) is beautiful. The author squashed space-time to 2D, explained the two postulates by geometric intuition and run a simulator on top of it. It is the 3blue1brown of SR!
0
0
5
Great paper exploring attention architecture for images! We encounter similar results in our latest work . Table 5 shows that keeping first layers conv, while using attention for last layers improve ResNet performance.
Further studies show that self-attention is the most useful in later layers while convolutions better capture lower-level features. Combining their will be an interesting research direction.
1
2
18
0
1
6
OMG a robot doing moonwalk and shuffle at the same time
Spot the robot dog dancing to UpTown Funk is simultaneously both terrifying and hilarious.
1K
23K
47K
0
0
5
exciting times
0
0
5
Formalization still needs humans, but now each human is 100X
Autoformalization with LLMs in Lean!
@zhangir_azerbay
and Edward Ayers built a chat interface to formalize natural language mathematics in Lean:
Very impressive work!
5
49
191
0
1
4
by being a creator of the transformer
I can't beleive I'm starting a company.... with David Ha
@hardmaru
!! .... in Tokyo?! How did I get here.
15
19
307
0
0
3
The NYT shares our work with nuanced perspectives from experts in different fields.
New York Times article on AlphaGeometry
Geometry was my favorite subject in high school because solving them requires many step of reasoning and planning. Geometry problems however have been difficult for AI to solve. Our Nature paper shows my team’s progress in Geometry
10
82
551
0
0
4
IMO 2022: China maxed out, S.Korea surpassed the US, and Vietnam came so close on the 4th position 🎉! Where are LLMs on this table 🤔
0
1
4
Finally some credit assignment
This article speaks many, I believe, hidden truths about Quoc Le on GoogleBrain & seq2seq. Personally, I have enjoyed working with Quoc who cares less about credit assignment but rather teamwork and long-term vision :)
2
91
438
0
0
4
a dream comes true!
Excited to announce Piano Genie, an intelligent controller that allows anyone to improvise on the piano! This was my internship project at
@GoogleMagenta
with
@iansimon
and
@sedielem
. Blog post with more information:
23
647
2K
0
0
3
@jasoncrawford
how about a field medalist in combinatorics?
This feels like a method that ought to be more generally applicable (as indeed the authors suggest). I have a few ideas of problems I'd be very interested for it to be tried out on and that seem to be of the right type.
2
4
72
0
0
3
skillful worldview:
i dont feel bad for ML PhDs who think "my job is ruined, i can't do anything outside a big lab" etc. i can think of >10 low hanging fruit things "someone should do" that im too busy for bc i have to "make revenue". academia is 100% unfettered from use value, yall have it so easy.
5
4
249
0
0
1
amazing & big congrats to the team!
Introducing Sora, our text-to-video model.
Sora can create videos of up to 60 seconds featuring highly detailed scenes, complex camera motion, and multiple characters with vibrant emotions.
Prompt: “Beautiful, snowy
10K
32K
138K
0
0
2
The next goalpost for AI has moved beyond my own reach...
0
0
2
Cool work applying RL on raw pixels!
Our latest work on continuous control from pixels by
@debidatta
, Jonathan Tompson,
@coreylynch
, and I will be presented at the MLPC workshop at
#ICRA2018
this afternoon in room M4. Paper:
2
23
75
0
0
1
tried this for translation a while back but didnt try hard enough -- i think the transformer needs way more registry & will improve on all tasks. also related to
@ChrSzegedy
"the transformer is so small" at 52:15 .
What I mean when I say “registers”: additional learnable tokens (like the [CLS]), but these ones are not used at output. No additional info at input, not used at output: these tokens could seem useless!
2
8
119
1
0
1
anyone who is a rationalist must accept this
0
0
1
failed to search for this song with bard/bing/chatgpt racking my brain with details from the MV.
then youtube found it in its first result :)
0
0
1
0
0
1
Tao grounded in formal environment
Nice, Terence Tao (Fields Medal 2006) found a bug in one of his papers using Lean 4.
16
328
2K
0
0
1
And on some tasks scaling has little effect. Or is it? Maybe the breakthrough point is just a few clicks ahead ... :)
Performance on some tasks improves smoothly with model scale, while on others there is sudden breakthrough performance at a critical scale.
2
12
117
0
0
1
NYU Courant pays PhD 54K? That's at least 1.5X the NYU CS rate then :)
Fellow PhDs! Let’s make our departments pay us a living wage!
6
59
179
2
0
1
one possible scenario: openai slack channel LTAO browsing X. because their method has nothing to do with q-learning nor rl. unlikely tho, because they are busy shipping q* into gpt5, not browsing X.
1
0
1
@finbarrtimbers
@krishnanrohit
maybe they trained on longer sequences where long term dep is important (e.g. large codebases)
1
0
1