[fix] Update bench command to use forward_prefill and forward_decode (#55) #58
Loading…
Reference in a new issue
No description provided.
Delete branch "fix/55-bench-command"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Summary
Fixes the
benchsubcommand which was callingmodel.forward()(which now returnserror.UseForwardPrefillDecode).Changes
bench_prefill: Usesmodel.forward_prefill()instead ofmodel.forward()bench_decode: Usesmodel.forward_prefill()for prompt, thenmodel.forward_decode()for token generationsync_linear_states_to_gpu()after prefill to match engine patternTest Results
zig buildcompiles cleanlyView command line instructions
Checkout
From your project repository, check out a new branch and test the changes.Merge
Merge the changes and update on Forgejo.Warning: The "Autodetect manual merge" setting is not enabled for this repository, you will have to mark this pull request as manually merged afterwards.