Skip to content
Server

Add --no-op-offload to improve -ot pp perf in MoE models like llama4 400B #13662

Sign in to view logs

Add --no-op-offload to improve -ot pp perf in MoE models like llama4 400B

Add --no-op-offload to improve -ot pp perf in MoE models like llama4 400B #13662

Triggered via pull request May 8, 2025 15:32
@hjc4869hjc4869
opened #13386
hjc4869:no_op_offload
Status Success
Total duration 1h 18m 7s
Artifacts –

server.yml

on: pull_request
server-windows
6m 27s
server-windows
Matrix: server
Fit to window
Zoom out
Zoom in