nodes/hao-ai-lab-FastVideo
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Implement training for hunyuan video

Browse Source
This commit is contained in:
Evelyn Zhang
2026-01-17 11:52:53 -08:00
committed by Will Lin
parent 57aea577c2
commit 5f8fb3cf5f
2 changed files with 15 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
fastvideo/tests/nightly/test_e2e_hunyuan_overfit_single_sample.py
+11 -10
View File
@@ -22,7 +22,7 @@ LOCAL_PREPROCESSED_DATA_DIR = Path(os.path.join(DATA_DIR, "cats_processed_t2v_hu
# training
NUM_GPUS_PER_NODE_TRAINING = "1"
NUM_GPUS_PER_NODE_TRAINING = "8"
TRAINING_ENTRY_FILE_PATH = "fastvideo/training/hunyuan_training_pipeline.py"
# New preprocessing pipeline creates files in training_dataset/worker_0/worker_0/
LOCAL_TRAINING_DATA_DIR = os.path.join(LOCAL_PREPROCESSED_DATA_DIR, "training_dataset", "worker_0", "worker_0")
@@ -112,9 +112,9 @@ def run_preprocessing():
"--preprocess.dataset_output_dir", str(LOCAL_PREPROCESSED_DATA_DIR),
"--preprocess.preprocess_video_batch_size", "1",
"--preprocess.dataloader_num_workers", "0",
"--preprocess.max_height", "384",
"--preprocess.max_width", "640",
"--preprocess.num_frames", "49",
"--preprocess.max_height", "480",
"--preprocess.max_width", "832",
"--preprocess.num_frames", "77",
"--preprocess.train_fps", "16",
"--preprocess.samples_per_file", "1",
"--preprocess.flush_frequency", "1",
@@ -161,9 +161,9 @@ def run_training():
"--training_cfg_rate", "0.0",
"--output_dir", str(LOCAL_OUTPUT_DIR),
"--tracker_project_name", "hunyuan_finetune_overfit_ci",
"--num_height", "384",
"--num_width", "640",
"--num_frames", "49",
"--num_height", "480",
"--num_width", "832",
"--num_frames", "81",
"--validation_guidance_scale", "1.0",
"--embedded_cfg_scale", "6.0",
"--num_euler_timesteps", "50",
@@ -182,11 +182,11 @@ def run_training():
def test_e2e_hunyuan_overfit_single_sample():
os.environ["WANDB_MODE"] = "online"
download_data()
#download_data()
run_preprocessing()
run_training()
reference_video_file = os.path.join(os.path.dirname(__file__), "reference_video_hunyuan_1_sample_v0.mp4")
reference_video_file = os.path.join(os.path.dirname(__file__), "reference_video_1_sample_v0.mp4")
print(f"reference_video_file: {reference_video_file}")
final_validation_video_file = os.path.join(LOCAL_OUTPUT_DIR, "validation_step_900_inference_steps_50_video_0.mp4")
print(f"final_validation_video_file: {final_validation_video_file}")
@@ -196,7 +196,7 @@ def test_e2e_hunyuan_overfit_single_sample():
assert os.path.exists(reference_video_file), f"Reference video not found at {reference_video_file}"
assert os.path.exists(final_validation_video_file), f"Validation video not found at {final_validation_video_file}"
# Compute SSIM
#Compute SSIM
mean_ssim, min_ssim, max_ssim = compute_video_ssim_torchvision(
reference_video_file,
final_validation_video_file,
@@ -211,5 +211,6 @@ def test_e2e_hunyuan_overfit_single_sample():
assert max_ssim > 0.5, f"Max SSIM is below 0.5: {max_ssim}"
if __name__ == "__main__":
test_e2e_hunyuan_overfit_single_sample()
fastvideo/tests/ssim/test_inference_similarity.py
+4
View File
@@ -19,12 +19,16 @@ if "A40" in device_name:
device_reference_folder = "A40" + device_reference_folder_suffix
elif "L40S" in device_name:
device_reference_folder = "L40S" + device_reference_folder_suffix
elif "H100" in device_name or "NVIDIA H100" in device_name:
device_reference_folder = "H100" + device_reference_folder_suffix
else:
# device_reference_folder = "L40S" + device_reference_folder_suffix
logger.warning(f"Unsupported device for ssim tests: {device_name}")
# raise ValueError(f"Unsupported device for ssim tests: {device_name}")
# Base parameters from the shell script
HUNYUAN_PARAMS = {
"num_gpus": 4,
"model_path": "FastVideo/FastHunyuan-diffusers",