Writing

• Exploring Mistral 7B: Unlocking High-Performance Language Models Oct 2023

  Why Mistral 7B punched well above its parameter count: the attention mechanisms behind it, how the benchmarks really compare, and what the licence lets you do.

• From Text to Sound: A Deep Dive into Text-to-Speech Generation and its Emerging Applications Sep 2023

  How text becomes speech: preprocessing, the acoustic model, the vocoder, and where TTS is heading beyond screen readers and voice assistants.

• The Bottlenecks of AGI: Why Can’t Machines Think Like Us… Yet? Sep 2023

  Today’s AI is to AGI what a pocket calculator is to a brain. The specific bottlenecks in the way: the engineering ones, not the hype ones.

• The Enigma of Memory: Do Large Language Models Truly Remember? Sep 2023

  Language models recall things without remembering anything. What memory means for a person, what it means for a model, and why conflating the two misleads.

• Hello World! Aug 2023

  What this blog is for, what is worth reading first, and why every post from 2023 and 2024 now opens by naming the thing it got wrong.