Siddharth Ancha
@siddancha
Followers
62
Following
108
Statuses
17
Postdoctoral Researcher @MIT_CSAIL @MIT | PhD from @SCSatCMU @CarnegieMellon | I work on Robotics 🤖
Cambridge, MA
Joined March 2010
@pika7ma "hunting dudes" might not be the most inclusive way to advertise PhD /postdoc positions 🙂
1
0
1
@jon_barron Maybe one difference is that inference time for diffusion models is usually fixed, whereas inference time is adaptive for o1 (thinks longer for harder problems)?
0
0
2
@allenzren Really cool work!!! The 2-layer MDP is neat and so are the best practices which seem to make DPPO work so well! I especially liked App. B.5 where you compare against evaluating the exact policy gradient using the DDPM flow-ODE. Although I wonder how FlowMatching might do here🤔
1
0
1
@ylecun @Frank37004246 Also, you can train reactive policies by learning to imitate MPC. RL is not the only way to learn fast policies.
1
0
3
@keenanisalive Hi @keenanisalive, really cool work! A slightly orthogonal question coming from someone outside the graphics community. What software/libraries do you use to create such amazing illustrations?
1
0
1
0
0
1
RT @AmandaCoston: Thrilled that our work on evaluating the justifiability of decision-making algorithms got a best paper award at SATML and…
0
12
0
RT @edwardhkennedy: Huge congrats to @AmandaCoston on her @Meta PhD research fellowship! Amanda's a joint student…
0
4
0
RT @mlcmublog: How can robots detect obstacles and navigate safely? Check out how we use a new controllable sensor called Programmable Lig…
0
3
0