Fix benchmark script (#32635)

* fix

* >= 0.3.0

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

.github/workflows/benchmark.yml

@@ -31,12 +31,12 @@ jobs:
         if: github.event_name == 'schedule'
         working-directory: /transformers
         run: |
-          python3 -m pip install optimum-benchmark>=0.2.0
+          python3 -m pip install optimum-benchmark>=0.3.0
           HF_TOKEN=${{ secrets.TRANSFORMERS_BENCHMARK_TOKEN }} python3 benchmark/benchmark.py --repo_id hf-internal-testing/benchmark_results --path_in_repo $(date +'%Y-%m-%d') --config-dir benchmark/config --config-name generation --commit=${{ github.sha }} backend.model=google/gemma-2b backend.cache_implementation=null,static backend.torch_compile=false,true --multirun
 
       - name: Benchmark (merged to main event)
         if: github.event_name == 'push' && github.ref_name == 'main'
         working-directory: /transformers
         run: |
-          python3 -m pip install optimum-benchmark>=0.2.0
+          python3 -m pip install optimum-benchmark>=0.3.0
           HF_TOKEN=${{ secrets.TRANSFORMERS_BENCHMARK_TOKEN }} python3 benchmark/benchmark.py --repo_id hf-internal-testing/benchmark_results_merge_event --path_in_repo $(date +'%Y-%m-%d') --config-dir benchmark/config --config-name generation --commit=${{ github.sha }} backend.model=google/gemma-2b backend.cache_implementation=null,static backend.torch_compile=false,true --multirun

benchmark/benchmark.py

@@ -101,7 +101,7 @@ def summarize(run_dir, metrics, expand_metrics=False):
         # post-processing of report: show a few selected/important metric
         for metric in metrics:
             keys = metric.split(".")
-            value = report
+            value = report.to_dict()
             current = metrics_values
             for key in keys:
                 # Avoid KeyError when a user's specified metric has typo.

setup.py

@@ -137,7 +137,7 @@ _deps = [
     "onnxruntime-tools>=1.4.2",
     "onnxruntime>=1.4.0",
     "opencv-python",
-    "optimum-benchmark>=0.2.0",
+    "optimum-benchmark>=0.3.0",
     "optuna",
     "optax>=0.0.8,<=0.1.4",
     "packaging>=20.0",

src/transformers/dependency_versions_table.py

@@ -43,7 +43,7 @@ deps = {
     "onnxruntime-tools": "onnxruntime-tools>=1.4.2",
     "onnxruntime": "onnxruntime>=1.4.0",
     "opencv-python": "opencv-python",
-    "optimum-benchmark": "optimum-benchmark>=0.2.0",
+    "optimum-benchmark": "optimum-benchmark>=0.3.0",
     "optuna": "optuna",
     "optax": "optax>=0.0.8,<=0.1.4",
     "packaging": "packaging>=20.0",