npu: read the token embedding table in place, one row per token, instead of materialising it (engine#1784) - #1795

npu: read the token embedding table in place, one row per token, instead of materialising it (engine#1784)#1795

bong-water-water-bong merged 1 commit into

main1bit-MONSTER/engine:mainfrom

1bit/1784-embed-mmap1bit-MONSTER/engine:1bit/1784-embed-mmapCopy head branch name to clipboard