Robotics @ OpenAI, CS PhD from UW, 🍕 Enthusiant, He/Him

San Francisco, CA
Joined February 2019
Adam Fishman 🐟🤖🐠 retweeted
🤔 Can we train one policy to control a wide range of robots, from drones to quadrupeds, navigators to bimanual manipulators, and more? 🦾Introducing CrossFormer: a single policy that can perform manipulation, navigation, aviation, and locomotion: crossformer-model.github.io
9
103
17
551
85,274
Adam Fishman 🐟🤖🐠 retweeted
Is it possible to have a single model for primitive actions that work on real robots 🤖 and unseen objects 🛩️? Introducing M2T2: Multi-task Masked Transformer at #CoRL23. M2T2 achieves 0-shot sim2real transfer in rearranging unseen objects in clutter. m2-t2.github.io 🧵👇
1
6
2
17
4,813
Adam Fishman 🐟🤖🐠 retweeted
🚀 Introducing our latest work, STOW, at #CoRL23! It aims at Segmenting and Tracking unseen Objects for Warehouse picking robots, through numerous items and discrete frames. Remarkable boost in grasping success rates from ~40% to ~70% across 82 trials! sites.google.com/view/stow-c… 🤖
6
1
10
5,858
Since we've been looking for more things to do, @emilymbender @mmitchell_ai @mcmillan_majora and I wrote a statement about the horrible "letter" on the AI apocalypse, the very first citation of which, was our #StochasticParrots paper. dair-institute.org/blog/lett…
35
757
191
1,564
802,317
Adam Fishman 🐟🤖🐠 retweeted
Okay, so that AI letter signed by lots of AI researchers calling for a "Pause [on] Giant AI Experiments"? It's just dripping with #Aihype. Here's a quick rundown. >>
30
445
171
1,297
519,439
Adam Fishman 🐟🤖🐠 retweeted
I finally find an explanation for why RL is needed for RLHF that satisfied me. It's actually like playing board games. The reward model can only judge a full answer and a "critic" is needed to efficiently improve the intermediate moves (earlier tokens in the answer) 1/4
12
162
10
1,261
215,409
Adam Fishman 🐟🤖🐠 retweeted
Hi, are you an undergrad at HBCU interested in AI research? We're hosting the 3rd summer research opportunity for HBCU students at BAIR 🥳bair.berkeley.edu/reu.html Paid, in-person program w/housing. Deadline 1/29/2023! Please share with your network 📢🙏 @berkeley_ai #HBCU
2
16
2
29
Adam Fishman 🐟🤖🐠 retweeted
All of twitter’s ML Ethics, transparency, and accountability team (except one). was laid off today. So much for that responsible ML initiative, huh?
26
111
16
752
Adam Fishman 🐟🤖🐠 retweeted
How can robots perform a wide variety of novel tasks from natural language? Execited to present Code as Policies - using language models to directly write robot policy code from language instructions. See paper, colabs, blog, and demos at code-as-policies.github.io long 🧵👇
16
145
38
651
Adam Fishman 🐟🤖🐠 retweeted
Excited to share our #NeurIPS2022 @NVIDIAAI work GET3D, a generative model that directly produces explicit textured 3D meshes with complex topology, rich geometric details, and high fidelity textures. #3D Project page: nv-tlabs.github.io/GET3D/
37
540
178
2,687
Adam Fishman 🐟🤖🐠 retweeted
If @PyTorch was printing the sizes, dtype and devices of all the tensors involved in an operation that failed, we would be getting AGI ten years earlier.
16
79
12
1,156
Adam Fishman 🐟🤖🐠 retweeted
Objectron a dataset of 15,000 annotated videos and 4M annotated images! GitHub github.com/google-research-d…
119
3
743
Adam Fishman 🐟🤖🐠 retweeted
Standing at #4 in our top 20 Robotic stories of 2020; @OpenAI abandons robotics research. The company will continue to invest in other areas where there is adequate data for training. Learn more:⤵️ bit.ly/3yrcUgt
2
2
Adam Fishman 🐟🤖🐠 retweeted
"No piece of art has ever emotionally affected me the way this robot arm piece has. It's programmed to try to contain the hydraulic fluid that’s constantly leaking out and required to keep itself running...
426
11,416
2,593
40,043