Argmax

Vahe Hagopian, Taka Hasegawa, Farrukh Rahman

A show where three machine learning enthusiasts talk about recent papers and developments in machine learning. Watch our video on YouTube https://www.youtube.com/@argmaxfm

Subscribe Listen in podnod RSS feed Show website

17 episodes
Updated Oct 8, 2024

Episodes17

Newest first

Oct 8, 2024 · 54 min
Mixture of Experts
In this episode we talk about the paper "Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer" by Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, Jeff Dean.
Sep 2, 2023 · 1 hr 2 min
LoRA
We talk about Low Rank Approximation for fine tuning Transformers. We are also on YouTube now! Check out the video here: https://youtu.be/lLzHr0VFi3Y
Mar 28, 2023 · 57 min
15: InstructGPT
In this episode we discuss the paper "Training language models to follow instructions with human feedback" by Ouyang et al (2022). We discuss the RLHF paradigm and how important RL is to tuning GPT.
Mar 17, 2023 · 49 min
14: Whisper
This week we talk about Whisper. It is a weakly supervised speech recognition model.
Mar 11, 2023 · 49 min
13: AlphaTensor
We talk about AlphaTensor, and how researchers were able to find a new algorithm for matrix multiplication.
Oct 25, 2022 · 54 min
12: SIRENs
In this episode we talked about "Implicit Neural Representations with Periodic Activation Functions" and the strength of periodic non-linearities.
Sep 30, 2022 · 48 min
11: CVPR Workshop on Autonomous Driving Keynote by Ashok Elluswamy, a Tesla engineer
In this episode we discuss this video: https://youtu.be/jPCV4GKX9Dw How Tesla approaches collision detection with novel methods.
Aug 23, 2022 · 54 min
10: Outracing champion Gran Turismo drivers with deep reinforcement learning
We discuss Sony AI's accomplishment of creating a novel AI agent that can beat professional racers in Gran Turismo. Some topics include: - The crafting of rewards to make the agent behave nicely - What is QR-SAC? - How to deal with "rare" experiences in the replay buffer Link to paper: https://www.nature.com/articles/s41586-021-04357-7
Jul 29, 2022 · 44 min
8: GATO (A Generalist Agent)
Today we talk about GATO, a multi-modal, multi-task, multi-embodiment generalist agent.
Jul 29, 2022 · 47 min
9: Heads-Up Limit Hold'em Poker Is Solved
Today we talk about recent AI advances in Poker; specifically the use of counterfactual regret minimization to solve the game of 2-player Limit Texas Hold'em.
Jun 14, 2022 · 30 min
7: Deep Unsupervised Learning Using Nonequilibrium Thermodynamics (Diffusion Models)
We start talking about diffusion models as a technique for generative deep learning.
Jun 6, 2022 · 1 hr 1 min
6: Deep Reinforcement Learning at the Edge of the Statistical Precipice
We discuss NeurIPS outstanding paper award winning paper, talking about important topics surrounding metrics and reproducibility.
Apr 26, 2022 · 42 min
5: QMIX
We talk about QMIX https://arxiv.org/abs/1803.11485 as an example of Deep Multi-agent RL.
Apr 6, 2022 · 55 min
4: Can Neural Nets Learn the Same Model Twice?
Todays paper: Can Neural Nets Learn the Same Model Twice? Investigating Reproducibility and Double Descent from the Decision Boundary Perspective (https://arxiv.org/pdf/2203.08124.pdf) Summary: A discussion of reproducibility and double descent through visualizations of decision boundaries. Highlights of the discussion: Relationship between model performance and reproducibility Which models are robust and reproducible How they calculate the various scores
Mar 21, 2022 · 44 min
3: VICReg
Todays paper: VICReg (https://arxiv.org/abs/2105.04906) Summary of the paper VICReg prevents representation collapse using a mixture of variance, invariance and covariance when calculating the loss. It does not require negative samples and achieves great performance on downstream tasks. Highlights of discussion The VICReg architecture (Figure 1) Sensitivity to hyperparameters (Table 7) Top 5 metric usefulness
Mar 7, 2022 · 53 min
2: data2vec
Todays paper: data2vec (https://arxiv.org/abs/2202.03555) Summary of the paper A multimodal SSL algorithm that predicts latent representation of different types of input. Highlights of discussion What are the motivations of SSL and multimodal How does the student teacher learning work? What are similarities and differences between ViT, BYOL, and Reinforcement Learning algorithms.
Feb 21, 2022 · 54 min
1: Reward is Enough
This is the first episode of Argmax! We talk about our motivations for doing a podcast, and what we hope listeners will get out of it. Todays paper: Reward is Enough Summary of the paper The authors present the Reward is Enough hypothesis: Intelligence, and its associated abilities, can be understood as subserving the maximisation of reward by an agent acting in its environment. Highlights of discussion High level overview of Reinforcement Learning How evolution can be encoded as a reward maximization problem What is the one reward signal we are trying to optimize?