2024-06-12
LLMs are distributed as huge files that contain compressed "weights". Weight is a probability of signal going from one node in the network to another.
LLMs need tools to run them. Most tools have some kind of <code>llama</code> reference in their name.
<a target="_blank" href="https://github.com/Mozilla-Ocho/llamafile"><code>llamafile</code></a> makes standalone LLM executables. It packs LLM model together with <code>llama.cpp</code> tool in one file, which you can download and run right away.

*2024-06-12*

**LLMs are distributed as huge files** that contain compressed "weights". Weight is a probability of signal going from one node in the network to another.

**LLMs need tools to run them**. Most tools have some kind of `llama` reference in their name.

[`llamafile`](https://github.com/Mozilla-Ocho/llamafile) **makes standalone LLM executables**. It packs LLM model together with `llama.cpp` tool in one file, which you can download and run right away.

Notes on LLMs