Eli Sennesh
@EliSennesh
Followers
788
Following
19K
Media
254
Statuses
6K
NHP ephys @VanderbiltU. Predictive coding, affective science. Abolish the value function! Thou shalt not make a machine to counterfeit a human mind.
Nashville, TN
Joined January 2024
PREPRINT THREAD🧵! Ever wanted to know how to ABOLISH THE VALUE FUNCTION, and why IT'S ALL TAXIS NAVIGATION? @mjdramstead and I wrote up some notes for y'all.
4
18
66
RT @EugeneVinitsky: Still in stealth but our team has grown to 20 and we're still hiring. If you're interested in joining the research fron….
0
16
0
We have got to come up with a term to replace "System 2" that didn't fall to the Replication Crisis in psychology!.
Excited to share Energy-Based Transformers (EBTs), which allows you to implement system 2 thinking in any modality! . EBTs formulate reasoning as an energy optimization problem, allowing models to internally think without complexities like CoT or multiple recurrent latents.
1
0
8
I like how they're admitting where the algorithmic hard part is and nonetheless focusing on making a scalable, expressive system for these problems.
@samuelekpe Yes, you can use more complex optimization algorithms to optimize the energy function (i.e. some of the graph search algorithms in .
0
0
7
RT @ZennaTavares: I'm in SF until Friday. Keen to meet people about @BasisOrg: automating robotics, city models. scientific discovery, col….
0
4
0
RT @xuanalogue: it's bc MCMC algorithms take a while to "burn in" before they converge to the true posterior distribution you're trying to….
0
3
0
RT @manorlaboratory: My lab’s 5-year NIH R01 grant, awarded to study gene therapy for hearing loss, was abruptly terminated. I want to shar….
0
2K
0
RT @marcelomattar: Thrilled to see our TinyRNN paper in @nature! We show how tiny RNNs predict choices of individual subjects accurately wh….
0
48
0
RT @du_yilun: Excited to share our workshop on Robotics World Modeling @corl_conf 2025!. World models have a huge range of applications in….
0
14
0
RT @KanakaRajanPhD: (6/7) A 4096-unit agent that remembers, plans & navigates risks gives a “window-sized” brain we can watch neuron-by-neu….
0
1
0
Finally, the ForageWorld paper!.
(2/7) Introducing ForageWorld: Each session spawns a large arena with lakes, predators & food patches that deplete over time. The AI agent must juggle hunger, thirst & fatigue in this virtual space. The agent can only "see" a small patch around itself, so no bird’s-eye view.
0
0
4
RT @CogCompNeuro: The 8th Cognitive Computational Neuroscience conference is coming closer and we are looking forward to welcoming the CCN….
0
5
0
RT @JoHenrich: New PNAS paper (Great work Ivan Kroupin): The cultural construction of executive function. By comparing cognitive developmen….
0
56
0
Today: NeurIPS reviews submitted.Tomorrow: the emails I owe a bunch of people.
1
0
6
RT @_TheTransmitter: The neurons that encode sequential information into working memory do not fire in that same order during recall, a fin….
0
14
0
RT @kareem_carr: something i really don't get about tech culture is they revere grinding. studying math, getting into the weeds, locking in….
0
33
0
RT @davidbau: The New England Mechanistic Interpretability Workshop, NEMI 2025 is August 22 in Boston. Talks, posters, meals, discussion. ….
0
9
0
RT @rao2z: What exactly is the alpha in suiting up to make yet another proprietary frontier model with the same architecture as everyone el….
0
4
0
RT @HenryYin19: NIH is slow to send funds for non-competing renewals. This is disrupting research in many labs, including mine. I was exp….
0
24
0