Stories by che_shr_cat

Make Softmax Great Again

Deep Learning Frameworks: The Fourth Pillar of Deep Learning Revolution

TextGrad: Automatic "Differentiation" via Text

Superconducting Supercomputers

Decoder-decoder architecture is coming

Chronos: Using Pretrained LLMs for Probabilistic Time Series Forecasting

Big Post About Big Context

Neural Network Diffusion

Thermodynamic AI is getting hotter

Training LLMs with AMD GPUs on Frontier Supercomputer

Beyond Chinchilla-Optimal Accounting for Inference in Language Model Scaling Law

Project CETI

GonzoML on Mamba and S6 (+previous post on S4)

Conway's Game of Life Is Omniperiodic

GonzoML on Gemini

Matryoshka Representation Learning

Mindstorms in Natural Language-Based Societies of Mind

The convolution empire strikes back

Sparse Universal Transformer

MemWalker: An alternative way for working with long documents using transformers

1
2
3
4