See you at #NeurIPS2022 My team there:
Learning and Control
Neural operators
Optimization
AI for Chemistry
Minedojo and VIMA: RL with language models
Neurosymbolic AI
Quantum+AI
Conversation
At #NeurIPS2022 you can also meet team:
Minedojo and VIMA: RL with language models
AI + Robotics
Language models, robustness and privacy
Multimodal AI
Diffusion models
Email/DM to talk about internships/full-time roles
4
4
69
Replying to
Large language models (LLMs) encode massive human common sense knowledge. Can we harness them for high-level task planning? Our #NeurIPS2022 Oral paper uses pre-trained LLMs as the policy backbone, leading to strong zero-shot generalizations to new tasks.
2
10
66
1
4
9
Quote Tweet
Large LM can sometimes give toxic generations
In #NeurIPS2022 we systematically understand and mitigate the toxicity issues from:
training corpus
model size (from 126M to 530B)
parameter efficiency
arxiv.org/abs/2202.04173
w/ @nvidia @IllinoisCS @Caltech @ASUShow this thread
1
2
Quote Tweet
Can we further improve the zero-shot generalization of foundation models (e.g., #CLIP)?
Our #NeurIPS 2022 paper introduces TPT, a Test-time Prompt Tuning framework that learns adaptive prompts for every single test sample.
Show this thread
0:18
12.6K views
1
1
3
Quote Tweet
Fast sampling of diffusion models. Only one model evaluation achieves SOTA! Check out poster at #NeurIPS22 SBM workshop on Friday 2:30pm-4pm at Room 293 - 294. @Kay12400259 @wn8_nie @ArashVahdat @Azizzadenesheli @nvidia @caltech twitter.com/_akhaliq/statu…
1
2
Quote Tweet
#Neurips2022 come to our "Learning Dissipative Dynamics in Chaotic Systems" on Wed at 11am, Hall J #740. We learn Kolmogorov Flow with Reynolds numbers up to 5000 and enforce dissipativity which is stable against O(10^4) perturbation.
@AnimaAnandkumar @Azizzadenesheli @NKovachki
Show this thread
GIF
1
3
Quote Tweet
Thanks for sharing @_akhaliq! Also checkout our oral session and poster at #NeurIPS2022 MLSys workshop!
nips.cc/Conferences/20
1
1
Quote Tweet
#NeurIPS2022 come to see our poster "Incremental Fourier Neural Operator" at AI4Science Workshop on Fri. (12pm). We incrementally augment frequency modes and data resolution during the training of FNO, which brings better generalization and less computational cost. twitter.com/ZongyiLiCaltec…
Show this thread
1
Quote Tweet
Come see a preview of our on-going work towards linear-scaling simulations of many-body quantum systems at the poster session of the #ai4science workshop at #neurips. Friday, 12-13PM, room 388. sharir.org/papers/vq-nqs
Show this thread
1
2
Quote Tweet
Tune in for our talk on Falcon, first model-based RL method in real-world extreme turbulent environments (Reynolds Num. 230000!), AI4Science workshop @ #Neurips2022, today @ 1:20pm. Falcon achieves 37% improvement over SOTA in stabilization of an airfoil using only 8 min of data!
Show this thread
GIF
1
3