Ask HN: How is a ML model represented on disk?
Genuinely curious and please point me to resources if this documented somewhere.
I come from a “everything on a computer is a file” perspective and am curious what a model persisted on disk looks like? Is it essentially a long list of key value pairs where the key is the node “ID” and value is its “current”weight (may be from its last training?)
Is there more to it? Look up the onnx file format.