user avatar
Eric Jang
@ericjang11
Joined January 2014
Posts
  • Pinned
    user avatar
    For the last few months I've been working on a from-scratch implementation of AlphaGo, a 2016 AI breakthrough that inspired me to get into deep learning. My casual understanding of AlphaGo was "search-augmented deep neural networks trained with self-play", but I wanted to go
    New blackboard lecture w @ericjang11 He walks through how to build AlphaGo from scratch, but with modern AI tools. Sometimes you understand the future better by stepping backward. AlphaGo is still the cleanest worked example of the primitives of intelligence: search, learning
    00:00
  • user avatar
    Looking forward to getting vaccinated every 4 nanoseconds
    Execute this code to debug COVID.
    covid_19 = True  while covid_19:  get_vaccinated   if indoors:  wear_mask()
  • user avatar
    The opening sentence goes so hard. This paper was 10 years ahead of its time.
  • user avatar
  • user avatar
    I look forward to a day where an artificial neural network can look at this sign and tell me if I can park here
  • user avatar
    Why is the market selling off Nvidia / compute stocks? The correct reaction to R1 breakthrough should be to buy *even more* compute.
  • user avatar
    Progress on NEO’s AI has been really fast of late. Here are some early clips of a generalist model we’re developing at @1x_tech. The following clips are 100% autonomous, running on a single set of neural network weights. First, a quiet little robot that picks up leaves and puts
    00:00
  • user avatar
    Every once in awhile a paper comes out that makes you breathe a sigh of relief that you don't publish in that field... arxiv.org/pdf/2003.08505… "Our results show that when hyperparameters are properly tuned via cross-validation, most methods perform similarly to one another"
  • user avatar
    We're inviting the first set of users to pre-order and experience NEO. This is a product that is early for its time. Some features are still in active development & polish. There will be mistakes. We will quickly learn from them, and use your early feedback to improve NEO for
    NEO The Home Robot Order Today
    00:00
  • user avatar
    The last part is left as an exercise to a neural network
  • user avatar
    ChatGPT is cool but I'm most impressed by how OpenAI is able to simultaneously serve a multi-billion parameter model to all the new users trying it out right now (including those coming from front page of HN). Kudos to whoever worked on the inference infra there
  • user avatar
    Instead of finding the perfect prompt for an LLM (let's think step by step), you can ask LLMs to critique their outputs and immediately fix their own mistakes. Here's a fun example:
  • user avatar
    This talk by @karpathy youtu.be/IHH47nZ7FZU has convinced me that Tesla is several years ahead of most CV labs in regards to pushing the limits of DL. Commonplace questions like "how do you do early stopping for a multi-task model?" are non-trivial when at scale.
  • user avatar
    Replying to @will__ye
    congrats! looking forward to see what you'll do next monday