For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python function

balanced_packing

`balanced_packing()`

max.pipelines.lib.eplb_rebalance.balanced_packing(weight, num_packs)

source

Pack n weighted objects into num_packs packs evenly.

Each pack contains exactly n / num_packs objects, with weights balanced as much as possible by greedy assignment.

Parameters:

weight (ndarray[tuple[Any, ...], dtype[float64]]) – [X, n] weight per item.
num_packs (int) – number of packs.

Returns:

[X, n] int64 — pack assignment per item. rank_in_pack: [X, n] int64 — position within the pack.

Return type:

pack_index