@@ -209,6 +209,8 @@ jobs:
209
209
210
210
# Convert HF checkpoint to ET via etLLM path
211
211
if [[ "$HF_MODEL_REPO" == meta-llama/* ]]; then
212
+ # The benchmark app replies on the _llm suffix to determine whether the model is a LLM or not
213
+ OUT_ET_MODEL_NAME=${OUT_ET_MODEL_NAME}_llm
212
214
# Llama models on Hugging Face
213
215
if [[ ${{ matrix.config }} == "llama3_spinquant" ]]; then
214
216
# SpinQuant
@@ -311,6 +313,7 @@ jobs:
311
313
ls -lh "${OUT_ET_MODEL_NAME}.pte"
312
314
fi
313
315
elif [[ "$HF_MODEL_REPO" == "Qwen/Qwen3-0.6B" ]]; then
316
+ OUT_ET_MODEL_NAME=${OUT_ET_MODEL_NAME}_llm
314
317
if [[ ${{ matrix.config }} == "et_xnnpack_custom_spda_kv_cache_8da4w" ]]; then
315
318
DOWNLOADED_PATH=$(bash .ci/scripts/download_hf_hub.sh --model_id "${HF_MODEL_REPO}" --subdir "." --files "tokenizer.json")
316
319
${CONDA_RUN} python -m examples.models.llama.export_llama \
@@ -367,12 +370,13 @@ jobs:
367
370
${CONDA_RUN} optimum-cli export executorch "${ARGS[@]}"
368
371
popd
369
372
373
+ # The benchmark app replies on the _llm suffix to determine whether the model is a LLM or not
374
+ OUT_ET_MODEL_NAME=${OUT_ET_MODEL_NAME}_llm
370
375
mv model.pte ${OUT_ET_MODEL_NAME}.pte
371
376
ls -lh "${OUT_ET_MODEL_NAME}.pte"
372
377
fi
373
378
374
- # zip -j model.zip ${OUT_ET_MODEL_NAME}.pte ${DOWNLOADED_PATH}/tokenizer.*
375
- zip -j model.zip ${OUT_ET_MODEL_NAME}.pte
379
+ zip -j model.zip ${OUT_ET_MODEL_NAME}.pte ${DOWNLOADED_PATH}/tokenizer.*
376
380
ls -lh model.zip
377
381
mkdir -p "${ARTIFACTS_DIR_NAME}"
378
382
mv model.zip "${ARTIFACTS_DIR_NAME}"
0 commit comments