Expand description
On-disk packed tensor layouts, independent of runtime sequence state.
Structs§
- Dense
Entry - Where a dense tensor lives inside
dense.bin. - Expert
Layout - Expert record layout inside
experts.bin. One record per (layer, expert): three packed 4-bit weight matrices followed by their scales and biases, padded to a page multiple so each record can be its own Metal buffer. - Manifest
- Ngram
Layout - N-gram table layout inside
ngram.bin: onerow_bytesrecord per hashed id holding the packed 4-bit row, then its scales, then its biases.