PL
@Avg_sapient
Followers
252
Following
1K
Media
34
Statuses
709
Autonomist | Robotics @lossfunk | RI @MIT-@NUSingapore , @TUDelft
State Space
Joined September 2022
The timing of this tweet is š¤š¤
"An autopsy of history would show that all great nations ultimately commit suicide" - attributed to historian Arnold Toynbee.
0
0
0
š
damn, spacing issues in tagging people š
tagging researchers correctly here: Aman Sharma @inceptmyth Chinmay Karkar @ChinmayKak Akshat Singh Jaswal @akshat_sj Ashish Baghel @nevashish Aman Sharma @inceptmyth Pranav Lakshmanan @Avg_sapient Ayush Nangia @vitransformer
0
0
5
Almost 5 NOTAMs in the north and west and now on the east , the game is getting very very interesting . Just near the chickens neck.
New Notam has been issued for an Indian Airforce Exercise in the Northeast.
0
0
1
That drop...is something. its possesed.
1
0
3
Saw this legendary piece of tech in my friends hostel room. Man runs unity on this , intel core i2 , 2gb ram . You can just do things.
2
0
5
If you think about it , seminconductor lasers are just inverted solar panels.
1
0
6
Yes lessgo robotics outta Lossfunk folks is getting to speed . The goats finally dropped. @Mankaran32 @pr0t0_01
Today we are launching 100 early pre-orders for Mr. Tensors @eyecandyrobots, an inexpensive opensource lifelike robot character that's cool enough toy for kids (and a human-robot-interaction + reinforcement learning research tool for academia and hobbyists). We are looking for
1
0
6
When you see so many posts about hacker houses , programs and grants and know that all the people selected are professional yappers but canāt prove itā¦
0
0
5
Additionally finding out my exams are in 2 different districts while an intense discussion on hidden representations and compositional generation is going on has gotta be a peak moment. Plus thanks for introducing me to new protein bars @dhruvtrehan9
after two hours of discussing llms predicting the future, good abstractions for continual learning, and if its possible to make questions too hard to answer - happy diwali from @lossfunk
0
0
5
X giving me almost daily reminders to get BACK into aerospace. Also why tf is ISRO not retweeting this. Cmon
Full bloom on the most powerful radar satellite ever builtāseen here with centimeter-level clarity. Vantor collected this image of the NISAR synthetic aperture radar (SAR) satellite after it deployed its giant circular radar antenna reflector in August. Developed jointly by
1
0
1
THAT IS ONE CLEAN FLAME . Bro...a methalox FFCC engine burning this clean and no weird flows even during gimbaling with such a clean design? Yeah idk what even to say here. Just this is art.
0
0
8
All rocket companies on one side and there is SpaceX flying starship like a fighter jet. At this point SpaceX hardware tests are a notch above any advertisement by any company on Earth to convince kids to pursue STEM. The keynotes of OAI , Claude , GDM are competing with this.
Starship is executing a banking maneuver that mimics the final approach it would take while returning to Starbase for a catch on a future mission
0
0
4
@lossfunk @retr0sushi_ @dhruvtrehan9 @ashishkr9311 @MajorTimbWlf21 Lmao tagged the wrong ashish @nevashish
1
0
4
3
0
14
Wtaf they topped opus 4.1? Yo Iām gonna be back after using this in bedrock .
Introducing Claude Sonnet 4.5āthe best coding model in the world. It's the strongest model for building complex agents. It's the best model at using computers. And it shows substantial gains on tests of reasoning and math.
0
0
2
Control theory is just a mathematician's stubborn attempt to condense the universe into their language, only to realize that they have to let their language learn on its own (ML/RL) and teach it to learn the world rather than teaching it to learn from their mind.
1
0
4
Rewatching squid games , holy fuck itās a master class in game theory. S3E3 is just peak parity matching .
0
0
2
Day 2 of Running S-Graphs on Custom data , random aura collection excercise. Fuck I love these aesthetics , plus mprocs being goated for data logging.
0
0
4
Heard it from the man himself fellas . I wasnāt bullish on LLMs earlier as well. Either some sort of WM based architecture or pure inductive reasoning based stuff is prolly the way to go.
@GaryMarcus @ylecun @demishassabis You were never alone, Gary, though you were the first to bite the bullet, to fight the good fight, and to make the argument well, again and again, for the limitations of LLMs. I salute you for this good service!
1
0
6