Kwangjin Oh retweeted
There's a lot of folks under the misunderstanding that it's now possible to run a 30B param LLM in <6GB, based on this GitHub discussion.
This is not the case. Understanding why gives us a chance to learn a lot of interesting stuff! π§΅
github.com/ggerganov/llama.cβ¦
Kwangjin Oh retweeted
Large Language Models 3.0
In the new issue of Ahead of AI, I am discussing what's next for LLMs and developments centered around parameter efficiency and multimodality.
It's also been a particularly strong month for open-source AI!
π magazine.sebastianraschka.coβ¦
Kwangjin Oh retweeted
"Build an AI to play Dino Run"
A tutorial to build a Reinforcement Learning model by @hellopaperspace
blog.paperspace.com/dino-runβ¦
Kwangjin Oh retweeted
iVQA: Inverse Visual Question Answering #CVPR2018
By Feng Liu et al.
Generate a question that corresponds to a given image and answer pair.
arxiv.org/abs/1710.03370
Kwangjin Oh retweeted
How to Implement a YOLO (v3) Object Detector from Scratch in PyTorch: Part 1 buff.ly/2L4WS2x
Kwangjin Oh retweeted
What is the Difference Between Deep Learning and βRegularβ Machine Learning? buff.ly/2KMODXX
Kwangjin Oh retweeted
Re-parameterize all the things! If you want to backprop thru samples from mixture, truncated, Gamma, Beta, Dirichlet, Student-t, or von Mises distributions, this paper has gotchu covered. Also faster than RSVI, another general reparameterization trick based on rejection sampling teapawt.net/BrundageBot/status/999β¦
Implicit Reparameterization Gradients. Michael Figurnov, Shakir Mohamed, and Andriy Mnih arxiv.org/abs/1805.08498
Kwangjin Oh retweeted
Monte Carlo Tree Search β Beginners Guide
int8.io/monte-carlo-tree-sea⦠#MachineLearning
Kwangjin Oh retweeted
Self-attention for GANs. No more problems with losing track of how many faces the generator has drawn on the dog. teapawt.net/gstsdn/status/99908347β¦
Kwangjin Oh retweeted
Implicit Reparameterization Gradients by myself, @shakir_za and Andriy Mnih: arxiv.org/abs/1805.08498. An extension of reparameterization for many distributions, including Gamma, Beta, Dirichlet and von Mises, that is faster and easier to use than the previous methods.
Kwangjin Oh retweeted
This website is amazing, linear algebra with interactive examples. Vectors, matrix, dot product, etc, cool resource for learning immersivemath.com/ila/index.β¦
Kwangjin Oh retweeted
μλ§κ° ννλ¬μ€μμ λΉμ²¨λμ΄μ λ€κ³ μ¨ νΈλμ νκΈ°μΈλ°γ
γ
γ
γ
γ
γ
γ
μμΉ¨μ λ΄κ° 건μ μ§ κ½μμ λλ Έλλ μ΄λ° γ
γ
γ
γ
γ
γ
γ
γ
γ
γ
μ΄κ±° μ΄λ»κ² λ€κ³ λ€λλΌκ³ γ
γ
γ
γ
γ
γ
γ
γ
γ
γ
μ€λ ν루μ μμμ§λ’°κ° λμ΄λ²λ ΈλΌγ
γ
γ
γ
γ
γ