gpt-oss: WIP llama

Browse Source

Model loads and runs (CPU only), but PPL is much to high
(~1500 for 1st batch vs ~200 in mainline).
Is it because of SWA, because of vocab, or did I introduce a bug somewhere?

...

This commit is contained in:

Iwan Kawrakow

2025-08-10 10:09:42 +03:00

parent e24a1d3eda

commit c69d04f324

2 changed files with 463 additions and 157 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

619

src/llama.cpp

File diff suppressed because it is too large Load Diff