fix environment problem

app.py

@@ -1,11 +1,34 @@
 import os
 from typing import List, Tuple
 
+import gradio as gr
+import spaces
+
 PWD = os.path.dirname(__file__)
 
 import subprocess
 
+# copy cudnn files
+subprocess.run("cp /usr/local/lib/python3.10/site-packages/nvidia/cudnn/include/*.h /usr/local/cuda/include", env={}, shell=True)
+subprocess.run("cp /usr/local/lib/python3.10/site-packages/nvidia/cudnn/lib/*.so* /usr/local/cuda/lib64", env={}, shell=True)
+
+# setup env
+os.environ["CUDA_HOME"] = "/usr/local/cuda"
+os.environ["LD_LIBRARY_PATH"] = "$CUDA_HOME/lib:$CUDA_HOME/lib64:$LD_LIBRARY_PATH"
+os.environ["PATH"] = "$CUDA_HOME/bin:/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:$PATH"
+
+# install packages
 subprocess.run("pip install flash-attn --no-build-isolation", env={"FLASH_ATTENTION_SKIP_CUDA_BUILD": "TRUE"}, shell=True)
+subprocess.run("pip install vllm==0.9.0", env={"VLLM_ATTENTION_BACKEND": "FLASHINFER"}, shell=True)
+subprocess.run(
+    "pip install transformer-engine[pytorch] --no-build-isolation",
+    env={"PATH": os.environ["PATH"], "LD_LIBRARY_PATH": os.environ["LD_LIBRARY_PATH"], "CUDA_HOME": os.environ["CUDA_HOME"]},
+    shell=True,
+)
+
+from test_environment import main as check_environment
+
+check_environment()
 
 try:
     import os
@@ -25,8 +48,8 @@ except Exception as e:
 # download checkpoints
 from download_checkpoints import main as download_checkpoints
 
-os.makedirs("./checkpoints", exist_ok=True)
-download_checkpoints(hf_token="", output_dir="./checkpoints", model="7b_av")
+os.makedirs(os.path.join(PWD, "checkpoints"), exist_ok=True)
+download_checkpoints(hf_token="", output_dir=os.path.join(PWD, "checkpoints"), model="7b_av")
 
 os.environ["TOKENIZERS_PARALLELISM"] = "false"  # Workaround to suppress MP warning
 
@@ -34,8 +57,6 @@ import copy
 import json
 import random
 from io import BytesIO
-
-import gradio as gr
 import torch
 from cosmos_transfer1.checkpoints import (
     BASE_7B_CHECKPOINT_AV_SAMPLE_PATH,
@@ -251,6 +272,7 @@ def inference(cfg, control_inputs) -> Tuple[List[str], List[str]]:
     return video_paths, prompt_paths
 
 
+@spaces.GPU()
 def generate_video(
     hdmap_video_input,
     lidar_video_input,
@@ -270,7 +292,7 @@ def generate_video(
             "hdmap": {"control_weight": 0.3, "input_control": hdmap_video_input},
             "lidar": {"control_weight": 0.7, "input_control": lidar_video_input},
         },
-        checkpoint_dir="./cosmos-transfer1/checkpoints",
+        checkpoint_dir=os.path.join(PWD, "checkpoints"),
         prompt=prompt,
         negative_prompt=negative_prompt,
         sigma_max=80,
@@ -300,19 +322,21 @@ with gr.Blocks() as demo:
             prompt_input = gr.Textbox(
                 label="Prompt",
                 lines=5,
-                value="A close-up shot captures a vibrant yellow scrubber vigorously working on a grimy plate, its bristles moving in circular motions to lift stubborn grease and food residue. The dish, once covered in remnants of a hearty meal, gradually reveals its original glossy surface. Suds form and bubble around the scrubber, creating a satisfying visual of cleanliness in progress. The sound of scrubbing fills the air, accompanied by the gentle clinking of the dish against the sink. As the scrubber continues its task, the dish transforms, gleaming under the bright kitchen lights, symbolizing the triumph of cleanliness over mess.",  # noqa: E501
+                # value="A close-up shot captures a vibrant yellow scrubber vigorously working on a grimy plate, its bristles moving in circular motions to lift stubborn grease and food residue. The dish, once covered in remnants of a hearty meal, gradually reveals its original glossy surface. Suds form and bubble around the scrubber, creating a satisfying visual of cleanliness in progress. The sound of scrubbing fills the air, accompanied by the gentle clinking of the dish against the sink. As the scrubber continues its task, the dish transforms, gleaming under the bright kitchen lights, symbolizing the triumph of cleanliness over mess.",  # noqa: E501
+                value="The video is captured from a camera mounted on a car. The camera is facing forward. The video showcases a scenic golden-hour drive through a suburban area, bathed in the warm, golden hues of the setting sun. The dashboard camera captures the play of light and shadow as the sun’s rays filter through the trees, casting elongated patterns onto the road. The streetlights remain off, as the golden glow of the late afternoon sun provides ample illumination. The two-lane road appears to shimmer under the soft light, while the concrete barrier on the left side of the road reflects subtle warm tones. The stone wall on the right, adorned with lush greenery, stands out vibrantly under the golden light, with the palm trees swaying gently in the evening breeze. Several parked vehicles, including white sedans and vans, are seen on the left side of the road, their surfaces reflecting the amber hues of the sunset. The trees, now highlighted in a golden halo, cast intricate shadows onto the pavement. Further ahead, houses with red-tiled roofs glow warmly in the fading light, standing out against the sky, which transitions from deep orange to soft pastel blue. As the vehicle continues, a white sedan is seen driving in the same lane, while a black sedan and a white van move further ahead. The road markings are crisp, and the entire setting radiates a peaceful, almost cinematic beauty. The golden light, combined with the quiet suburban landscape, creates an atmosphere of tranquility and warmth, making for a mesmerizing and soothing drive.",  # noqa: E501
                 placeholder="Enter your descriptive prompt here...",
             )
 
             negative_prompt_input = gr.Textbox(
                 label="Negative Prompt",
                 lines=3,
-                value="The video captures a series of frames showing ugly scenes, static with no motion, motion blur, over-saturation, shaky footage, low resolution, grainy texture, pixelated images, poorly lit areas, underexposed and overexposed scenes, poor color balance, washed out colors, choppy sequences, jerky movements, low frame rate, artifacting, color banding, unnatural transitions, outdated special effects, fake elements, unconvincing visuals, poorly edited content, jump cuts, visual noise, and flickering. Overall, the video is of poor quality.",  # noqa: E501
+                # value="The video captures a series of frames showing ugly scenes, static with no motion, motion blur, over-saturation, shaky footage, low resolution, grainy texture, pixelated images, poorly lit areas, underexposed and overexposed scenes, poor color balance, washed out colors, choppy sequences, jerky movements, low frame rate, artifacting, color banding, unnatural transitions, outdated special effects, fake elements, unconvincing visuals, poorly edited content, jump cuts, visual noise, and flickering. Overall, the video is of poor quality.",  # noqa: E501
+                value="The video captures a game playing, with bad crappy graphics and cartoonish frames. It represents a recording of old outdated games. The lighting looks very fake. The textures are very raw and basic. The geometries are very primitive. The images are very pixelated and of poor CG quality. There are many subtitles in the footage. Overall, the video is unrealistic at all.",  # noqa: E501
                 placeholder="Enter what you DON'T want to see in the image...",
             )
 
             with gr.Row():
-                randomize_seed_checkbox = gr.Checkbox(label="Randomize Seed", value=True)
+                randomize_seed_checkbox = gr.Checkbox(label="Randomize Seed", value=False)
                 seed_input = gr.Slider(minimum=0, maximum=1000000, value=1, step=1, label="Seed")
 
             generate_button = gr.Button("Generate Image")

cosmos_transfer1/diffusion/config/transfer/config.py

@@ -0,0 +1,57 @@
+# 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.
+
+from typing import Any, List
+
+import attrs
+
+from cosmos_transfer1.diffusion.config.transfer.model import CtrlModelConfig
+from cosmos_transfer1.diffusion.config.transfer.registry import register_configs
+from cosmos_transfer1.diffusion.model.model_ctrl import VideoDiffusionModelWithCtrl
+from cosmos_transfer1.utils import config
+from cosmos_transfer1.utils.config_helper import import_all_modules_from_package
+from cosmos_transfer1.utils.lazy_config import PLACEHOLDER
+from cosmos_transfer1.utils.lazy_config import LazyCall as L
+from cosmos_transfer1.utils.lazy_config import LazyDict
+
+
+@attrs.define(slots=False)
+class Config(config.Config):
+    # default config groups that will be used unless overwritten
+    # see config groups in registry.py
+    defaults: List[Any] = attrs.field(
+        factory=lambda: [
+            "_self_",
+            {"net": None},
+            {"net_ctrl": None},
+            {"hint_key": "control_input_edge"},
+            {"conditioner": "ctrlnet_add_fps_image_size_padding_mask"},
+            {"tokenizer": "vae1"},
+            {"experiment": None},
+        ]
+    )
+    model_obj: LazyDict = L(VideoDiffusionModelWithCtrl)(
+        config=PLACEHOLDER,
+    )
+
+
+def make_config():
+    c = Config(
+        model=CtrlModelConfig(),
+    )
+    register_configs()
+
+    import_all_modules_from_package("cosmos_transfer1.diffusion.config.inference")
+    return c

download_checkpoints.py

@@ -114,6 +114,7 @@ def main(hf_token: str = os.environ.get("HF_TOKEN"), output_dir: str = "./checkp
                 checkpoint_vars.append(obj)
 
     print(f"Found {len(checkpoint_vars)} checkpoints to download")
+    print(checkpoint_vars)
 
     # Download each checkpoint
     for checkpoint in checkpoint_vars:

requirements.txt

@@ -1,10 +1,15 @@
-git+https://github.com/huggingface/diffusers.git
-transformers
-accelerate
-sentencepiece
-safetensors
-torchvision
-git+https://github.com/yiyixuxu/cosmos-guardrail.git
-peft
+# essentials
+ninja
 
+# torch
+# torch==2.7.0
+# torchvision==0.22.0
+# torch==2.5.1
+# torchvision==0.20.1
+
+# cosmos-transfer1
 git+https://github.com/nvidia-cosmos/cosmos-transfer1
+
+# other packages
+https://download.pytorch.org/whl/cu128/flashinfer/flashinfer_python-0.2.5%2Bcu128torch2.7-cp38-abi3-linux_x86_64.whl
+decord==0.6.0

test_environment.py

@@ -0,0 +1,71 @@
+# 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 importlib
+import os
+import sys
+
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--training",
+        action="store_true",
+        help="Whether to check training-specific dependencies",
+    )
+    return parser.parse_args()
+
+
+def check_packages(package_list):
+    global all_success
+    for package in package_list:
+        try:
+            _ = importlib.import_module(package)
+        except Exception:
+            print(f"\033[91m[ERROR]\033[0m Package not successfully imported: \033[93m{package}\033[0m")
+            all_success = False
+        else:
+            print(f"\033[92m[SUCCESS]\033[0m {package} found")
+
+
+def main():
+    args = parse_args()
+
+    if not (sys.version_info.major == 3 and sys.version_info.minor >= 10):
+        detected = f"{sys.version_info.major}.{sys.version_info.minor}.{sys.version_info.micro}"
+        print(f"\033[91m[ERROR]\033[0m Python 3.10+ is required. You have: \033[93m{detected}\033[0m")
+        sys.exit(1)
+
+    if "CONDA_PREFIX" not in os.environ:
+        print(
+            "\033[93m[WARNING]\033[0m CONDA_PREFIX is not set. When manually installed, Cosmos should run under the cosmos-transfer1 conda environment (see INSTALL.md). This warning can be ignored when running in the container."
+        )
+
+    print("Attempting to import critical packages...")
+
+    packages = ["torch", "torchvision", "transformers", "megatron.core", "transformer_engine", "vllm", "pandas"]
+    packages_training = [
+        "apex.multi_tensor_apply",
+    ]
+    all_success = True
+
+    check_packages(packages)
+    if args.training:
+        check_packages(packages_training)
+
+    if all_success:
+        print("-----------------------------------------------------------")
+        print("\033[92m[SUCCESS]\033[0m Cosmos environment setup is successful!")