update whisper dataset folder structure

Author: anandhu-eng

script/app-mlperf-inference-mlcommons-python/customize.py

@@ -574,7 +574,7 @@ def get_run_cmd_reference(
         cmd = f"""{x}{env['MLC_PYTHON_BIN_WITH_PATH']}{x} reference_mlperf.py \
             --scenario {tmp_scenario} \
             --dataset_dir {x}{env['MLC_DATASET_WHISPER_PATH']}{x} \
-            --manifest {x}{os.path.join(env['MLC_DATASET_WHISPER_PATH'], "dev-all-repack.json")}{x} \
+            --manifest {x}{os.path.join(env['MLC_DATASET_WHISPER_PATH'], "data", "dev-all-repack.json")}{x} \
             --log_dir {x}{env['MLC_MLPERF_OUTPUT_DIR']}{x} \
             --model-path {x}{env['MLC_ML_MODEL_WHISPER_PATH']}{x}"""
 

script/get-dataset-whisper/customize.py

@@ -14,7 +14,6 @@ def preprocess(i):
     if env.get('MLC_DATASET_WHISPER_PATH', '') != '':
         return {'return': 0}
 
-    print(env.get('MLC_TMP_DATASET_TYPE', ''))
     if env.get('MLC_TMP_DATASET_TYPE', '') == "preprocessed":
         env['MLC_TMP_REQUIRE_DOWNLOAD'] = "yes"
     else:
@@ -42,4 +41,31 @@ def postprocess(i):
 
     env = i['env']
 
+    if env.get('MLC_TMP_DATASET_TYPE', '') != "preprocessed":
+        cwd = env.get('MLC_OUTDIRNAME', os.getcwd())
+        data_dir = os.path.join(cwd, 'data')
+        env['MLC_DATASET_WHISPER_PATH'] = data_dir
+    else:
+        # copy files to data folder
+        tmp_src_dir = env["MLC_DATASET_WHISPER_PATH"]
+        tmp_dest_dir = os.path.join(tmp_src_dir, "data")
+        
+        os.makedirs(tmp_dest_dir, exist_ok=True)
+
+        items_to_copy = [
+            "LibriSpeech",
+            "dev-all",
+            "dev-all-repack",
+            "dev-all-repack.json"
+        ]
+
+        for item in items_to_copy:
+            src_path = os.path.join(tmp_src_dir, item)
+            dst_path = os.path.join(tmp_dest_dir, item)
+            if os.path.isdir(src_path):
+                shutil.copytree(src_path, dst_path, dirs_exist_ok=True)
+            elif os.path.isfile(src_path):
+                shutil.copy2(src_path, dst_path)
+
+
     return {'return': 0}