Skip to content

Commit 6a96e09

Browse files
mergennachinfacebook-github-bot
authored andcommitted
OSS CI: Test our endorsed llama path
Summary: Currently, in OSS CI, we are testing xnnpack path, PT2E PTQ path. However, I realized that we don't actually have a test to exercise what's in the README file. Even for stories, let's test this path. I tested locally and it works. bypass-github-export-checks bypass-github-pytorch-ci-checks bypass-github-executorch-ci-checks Reviewed By: lucylq Differential Revision: D55822681 fbshipit-source-id: 794ef7c7c66ffe1ffb3d6971cea5c5dd5c736a8c
1 parent ba7bbe8 commit 6a96e09

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

.ci/scripts/test_llama.sh

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -118,7 +118,7 @@ EXPORTED_MODEL_NAME="${EXPORTED_MODEL_NAME}.pte"
118118
echo "Exporting ${EXPORTED_MODEL_NAME}"
119119
EXPORT_ARGS="-c stories110M.pt -p ${PARAMS} -d ${DTYPE} -n ${EXPORTED_MODEL_NAME}"
120120
if [[ "${MODE}" == "xnnpack" ]]; then
121-
EXPORT_ARGS="${EXPORT_ARGS} --pt2e_quantize xnnpack_dynamic"
121+
EXPORT_ARGS="${EXPORT_ARGS} -kv --use_sdpa_with_kv_cache -X -qmode 8da4w -G 128"
122122
fi
123123
$PYTHON_EXECUTABLE -m examples.models.llama2.export_llama ${EXPORT_ARGS}
124124

0 commit comments

Comments
 (0)