edit data preprocess script to avoid downloading cosmos2.5 repo

Author: Ting-Yun

examples/cosmos/create_prompts_for_gr1_dataset.py

@@ -0,0 +1,62 @@
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import argparse
+import os
+
+from tqdm import tqdm
+
+"""example command
+python -m scripts.create_prompts_for_gr1_dataset --dataset_path datasets/benchmark_train/gr1
+"""
+
+
+def parse_args() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser(description="Create text prompts for GR1 dataset")
+    parser.add_argument(
+        "--dataset_path", type=str, default="datasets/benchmark_train/gr1", help="Root path to the dataset"
+    )
+    parser.add_argument(
+        "--prompt_prefix", type=str, default="The robot arm is performing a task. ", help="Prefix of the prompt"
+    )
+    parser.add_argument(
+        "--meta_csv", type=str, default="datasets/benchmark_train/gr1/metadata.csv", help="Metadata csv file"
+    )
+    return parser.parse_args()
+
+
+def main(args) -> None:
+    meta_csv = args.meta_csv
+    meta_lines = open(meta_csv).readlines()[1:]
+    meta_txt_dir = os.path.join(args.dataset_path, "metas")
+    os.makedirs(meta_txt_dir, exist_ok=True)
+
+    for meta_line in tqdm(meta_lines):
+        video_filename, prompt = meta_line.split(",", 1)
+        prompt = prompt.strip("\n")
+        if prompt.startswith('"') and prompt.endswith('"'):
+            # Remove the quotes
+            prompt = prompt[1:-1]
+        prompt = args.prompt_prefix + prompt
+        meta_txt_filename = os.path.join(meta_txt_dir, os.path.basename(video_filename).replace(".mp4", ".txt"))
+        with open(meta_txt_filename, "w") as fp:
+            fp.write(prompt)
+
+        print(f"encoding prompt: {prompt}")
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    main(args)

examples/cosmos/download_and_preprocess_datasets.sh

@@ -3,15 +3,12 @@ train_dir=$dataset_dir/train
 test_dir=$dataset_dir/test
 
 # Download and Preprocess Training Dataset
-git clone https://github.com/nvidia-cosmos/cosmos-predict2.5.git
-cd cosmos-predict2.5
-
 hf download nvidia/GR1-100 --repo-type dataset --local-dir datasets/benchmark_train/hf_gr1/ && \
 mkdir -p datasets/benchmark_train/gr1/videos && \
 mv datasets/benchmark_train/hf_gr1/gr1/*mp4 datasets/benchmark_train/gr1/videos && \
 mv datasets/benchmark_train/hf_gr1/metadata.csv datasets/benchmark_train/gr1/
 
-python -m scripts.create_prompts_for_gr1_dataset --dataset_path datasets/benchmark_train/gr1
+python create_prompts_for_gr1_dataset.py --dataset_path datasets/benchmark_train/gr1
 
 # Download Eval Dataset
 hf download nvidia/EVAL-175 --repo-type dataset --local-dir dream_gen_benchmark
@@ -23,5 +20,3 @@ mv datasets/benchmark_train/gr1 $train_dir
 mv dream_gen_benchmark/gr1_object $test_dir
 echo Download training data to $train_dir
 echo Download test data to $test_dir
-mv $dataset_dir ..
-cd ..