starflow / app.py
leoeric's picture
Improve checkpoint caching check and set logdir for easier output finding
02a442c
raw
history blame
15.2 kB
"""
Hugging Face Space for STARFlow
Text-to-Image and Text-to-Video Generation
This app allows you to run STARFlow inference on Hugging Face GPU infrastructure.
"""
import os
import gradio as gr
import torch
import subprocess
import pathlib
from pathlib import Path
# Fix OpenMP warning
os.environ['OMP_NUM_THREADS'] = '1'
# Try to import huggingface_hub for downloading checkpoints
try:
from huggingface_hub import hf_hub_download
HF_HUB_AVAILABLE = True
except ImportError:
HF_HUB_AVAILABLE = False
print("⚠️ huggingface_hub not available. Install with: pip install huggingface_hub")
# Check if running on Hugging Face Spaces
HF_SPACE = os.environ.get("SPACE_ID") is not None
# Default checkpoint paths (if uploaded to Space Files)
DEFAULT_IMAGE_CHECKPOINT = "ckpts/starflow_3B_t2i_256x256.pth"
DEFAULT_VIDEO_CHECKPOINT = "ckpts/starflow-v_7B_t2v_caus_480p_v3.pth"
# Model Hub repositories (if using Hugging Face Model Hub)
# Set these to your Model Hub repo IDs if you upload checkpoints there
# Format: "username/repo-name"
IMAGE_CHECKPOINT_REPO = "GlobalStudio/starflow-3b-checkpoint" # Update this after creating Model Hub repo
VIDEO_CHECKPOINT_REPO = "GlobalStudio/starflow-v-7b-checkpoint" # Update this after creating Model Hub repo
def get_checkpoint_path(checkpoint_file, default_local_path, repo_id=None, filename=None):
"""Get checkpoint path, downloading from Hub if needed."""
# If user uploaded a file, use it
if checkpoint_file is not None:
if hasattr(checkpoint_file, 'name'):
return checkpoint_file.name
return str(checkpoint_file)
# Try local path first
if os.path.exists(default_local_path):
return default_local_path
# Try downloading from Model Hub if configured
if repo_id and filename and HF_HUB_AVAILABLE:
try:
# Use /workspace if available (persistent), otherwise /tmp
cache_dir = "/workspace/checkpoints" if os.path.exists("/workspace") else "/tmp/checkpoints"
os.makedirs(cache_dir, exist_ok=True)
# Check if already downloaded
possible_path = os.path.join(cache_dir, "models--" + repo_id.replace("/", "--"), "snapshots", "*", filename)
import glob
existing = glob.glob(possible_path)
if existing:
checkpoint_path = existing[0]
print(f"✅ Using cached checkpoint: {checkpoint_path}")
return checkpoint_path
# Download with progress tracking
import time
start_time = time.time()
print(f"📥 Downloading checkpoint from {repo_id}...")
print(f"File size: ~15.5 GB - This may take 10-30 minutes")
checkpoint_path = hf_hub_download(
repo_id=repo_id,
filename=filename,
cache_dir=cache_dir,
local_files_only=False,
resume_download=True, # Resume if interrupted
)
elapsed = time.time() - start_time
print(f"✅ Download completed in {elapsed/60:.1f} minutes")
print(f"✅ Checkpoint at: {checkpoint_path}")
return checkpoint_path
except Exception as e:
error_detail = str(e)
if "404" in error_detail or "not found" in error_detail.lower():
return None, f"Checkpoint not found in Model Hub.\n\nPlease verify:\n1. Repo exists: https://huggingface.co/{repo_id}\n2. File exists: {filename}\n3. Repo is Public (not Private)\n\nError: {error_detail}"
return None, f"Error downloading checkpoint: {error_detail}\n\nThis may take 10-30 minutes for a 14GB file. Please wait or check your internet connection."
# No checkpoint found
return None, f"Checkpoint not found. Please upload a checkpoint file or configure Model Hub repository."
# Verify CUDA availability (will be True on HF Spaces with GPU hardware)
if torch.cuda.is_available():
print(f"✅ CUDA available! Device: {torch.cuda.get_device_name(0)}")
print(f" CUDA Version: {torch.version.cuda}")
print(f" PyTorch Version: {torch.__version__}")
else:
print("⚠️ CUDA not available. Make sure GPU hardware is selected in Space settings.")
def generate_image(prompt, aspect_ratio, cfg, seed, checkpoint_file, config_path):
"""Generate image from text prompt."""
# Get checkpoint path (from upload, local, or Model Hub)
result = get_checkpoint_path(
checkpoint_file,
DEFAULT_IMAGE_CHECKPOINT,
IMAGE_CHECKPOINT_REPO,
"starflow_3B_t2i_256x256.pth"
)
if isinstance(result, tuple) and result[0] is None:
return None, result[1] # Error message
checkpoint_path = result
# Show status
status_msg = f"Using checkpoint: {checkpoint_path}\n"
if not os.path.exists(checkpoint_path):
return None, f"Error: Checkpoint file not found at {checkpoint_path}.\n\nPlease verify:\n1. Model Hub repo exists: {IMAGE_CHECKPOINT_REPO}\n2. File name matches: starflow_3B_t2i_256x256.pth\n3. Repo is Public (not Private)"
if not config_path or not os.path.exists(config_path):
return None, "Error: Config file not found. Please ensure config file exists."
status_msg += "Starting image generation...\n"
status_msg += "This may take 1-3 minutes for first run (model loading).\n"
try:
# Create output directory
output_dir = Path("outputs")
output_dir.mkdir(exist_ok=True)
# Run sampling command
# Set logdir to outputs directory for easier file finding
cmd = [
"python", "sample.py",
"--model_config_path", config_path,
"--checkpoint_path", checkpoint_path,
"--caption", prompt,
"--sample_batch_size", "1",
"--cfg", str(cfg),
"--aspect_ratio", aspect_ratio,
"--seed", str(seed),
"--save_folder", "1",
"--finetuned_vae", "none",
"--jacobi", "1",
"--jacobi_th", "0.001",
"--jacobi_block_size", "16",
"--logdir", str(output_dir) # Set logdir to outputs
]
status_msg += "Running generation...\n"
status_msg += "Note: First run includes checkpoint download (~10-20 min) and model loading (~2-5 min).\n"
# Run with timeout (45 minutes max - allows for download + generation)
result = subprocess.run(cmd, capture_output=True, text=True, cwd=os.getcwd(), timeout=2700)
if result.returncode != 0:
error_msg = f"Error during generation:\n{result.stderr}\n\nStdout:\n{result.stdout}"
return None, error_msg
status_msg += "Generation complete. Looking for output...\n"
# Find the generated image
# The sample.py script saves to logdir/model_name/...
# We need to find the most recent output
output_files = list(output_dir.glob("**/*.png")) + list(output_dir.glob("**/*.jpg"))
if output_files:
latest_file = max(output_files, key=lambda p: p.stat().st_mtime)
return str(latest_file), status_msg + "✅ Success! Image generated."
else:
return None, status_msg + f"Error: Generated image not found in {output_dir}. Check stdout:\n{result.stdout}"
except Exception as e:
return None, f"Error: {str(e)}"
def generate_video(prompt, aspect_ratio, cfg, seed, target_length, checkpoint_file, config_path, input_image):
"""Generate video from text prompt."""
# Get checkpoint path (from upload, local, or Model Hub)
result = get_checkpoint_path(
checkpoint_file,
DEFAULT_VIDEO_CHECKPOINT,
VIDEO_CHECKPOINT_REPO,
"starflow-v_7B_t2v_caus_480p_v3.pth"
)
if isinstance(result, tuple) and result[0] is None:
return None, result[1] # Error message
checkpoint_path = result
if not os.path.exists(checkpoint_path):
return None, f"Error: Checkpoint file not found at {checkpoint_path}."
if not config_path or not os.path.exists(config_path):
return None, "Error: Config file not found. Please ensure config file exists."
# Handle input image
input_image_path = None
if input_image is not None:
if hasattr(input_image, 'name'):
input_image_path = input_image.name
else:
input_image_path = str(input_image)
try:
# Create output directory
output_dir = Path("outputs")
output_dir.mkdir(exist_ok=True)
# Run sampling command
cmd = [
"python", "sample.py",
"--model_config_path", config_path,
"--checkpoint_path", checkpoint_path,
"--caption", prompt,
"--sample_batch_size", "1",
"--cfg", str(cfg),
"--aspect_ratio", aspect_ratio,
"--seed", str(seed),
"--out_fps", "16",
"--save_folder", "1",
"--jacobi", "1",
"--jacobi_th", "0.001",
"--finetuned_vae", "none",
"--disable_learnable_denoiser", "0",
"--jacobi_block_size", "32",
"--target_length", str(target_length)
]
if input_image_path and os.path.exists(input_image_path):
cmd.extend(["--input_image", input_image_path])
else:
cmd.extend(["--input_image", "none"])
result = subprocess.run(cmd, capture_output=True, text=True, cwd=os.getcwd())
if result.returncode != 0:
return None, f"Error: {result.stderr}"
# Find the generated video
output_files = list(output_dir.glob("**/*.mp4")) + list(output_dir.glob("**/*.gif"))
if output_files:
latest_file = max(output_files, key=lambda p: p.stat().st_mtime)
return str(latest_file), "Success! Video generated."
else:
return None, "Error: Generated video not found."
except Exception as e:
return None, f"Error: {str(e)}"
# Create Gradio interface
with gr.Blocks(title="STARFlow - Text-to-Image & Video Generation") as demo:
gr.Markdown("""
# STARFlow: Scalable Transformer Auto-Regressive Flow
Generate high-quality images and videos from text prompts using STARFlow models.
**Checkpoints are automatically downloaded from Model Hub on first use.**
""")
with gr.Tabs():
with gr.Tab("Text-to-Image"):
with gr.Row():
with gr.Column():
image_prompt = gr.Textbox(
label="Prompt",
placeholder="a film still of a cat playing piano",
lines=3
)
# Checkpoint upload hidden - using Model Hub instead
image_checkpoint = gr.File(
label="Model Checkpoint (.pth file) - Optional if already uploaded to Space",
file_types=[".pth"],
visible=False # Hidden - using Model Hub
)
image_config = gr.Textbox(
label="Config Path",
value="configs/starflow_3B_t2i_256x256.yaml",
placeholder="configs/starflow_3B_t2i_256x256.yaml"
)
image_aspect = gr.Dropdown(
choices=["1:1", "2:3", "3:2", "16:9", "9:16", "4:5", "5:4"],
value="1:1",
label="Aspect Ratio"
)
image_cfg = gr.Slider(1.0, 10.0, value=3.6, step=0.1, label="CFG Scale")
image_seed = gr.Number(value=999, label="Seed", precision=0)
image_btn = gr.Button("Generate Image", variant="primary")
with gr.Column():
image_output = gr.Image(label="Generated Image")
image_status = gr.Textbox(label="Status", interactive=False)
image_btn.click(
fn=generate_image,
inputs=[image_prompt, image_aspect, image_cfg, image_seed, image_checkpoint, image_config],
outputs=[image_output, image_status],
show_progress=True
)
with gr.Tab("Text-to-Video"):
with gr.Row():
with gr.Column():
video_prompt = gr.Textbox(
label="Prompt",
placeholder="a corgi dog looks at the camera",
lines=3
)
# Checkpoint upload hidden - using Model Hub instead
video_checkpoint = gr.File(
label="Model Checkpoint (.pth file) - Optional if already uploaded to Space",
file_types=[".pth"],
visible=False # Hidden - using Model Hub
)
video_config = gr.Textbox(
label="Config Path",
value="configs/starflow-v_7B_t2v_caus_480p.yaml",
placeholder="configs/starflow-v_7B_t2v_caus_480p.yaml"
)
video_aspect = gr.Dropdown(
choices=["16:9", "1:1", "4:3"],
value="16:9",
label="Aspect Ratio"
)
video_cfg = gr.Slider(1.0, 10.0, value=3.5, step=0.1, label="CFG Scale")
video_seed = gr.Number(value=99, label="Seed", precision=0)
video_length = gr.Slider(81, 481, value=81, step=80, label="Target Length (frames)")
video_input_image = gr.File(
label="Input Image (optional, for image-to-video)",
file_types=["image"]
)
video_btn = gr.Button("Generate Video", variant="primary")
with gr.Column():
video_output = gr.Video(label="Generated Video")
video_status = gr.Textbox(label="Status", interactive=False)
video_btn.click(
fn=generate_video,
inputs=[video_prompt, video_aspect, video_cfg, video_seed, video_length,
video_checkpoint, video_config, video_input_image],
outputs=[video_output, video_status],
show_progress=True
)
if __name__ == "__main__":
# Password protection - users don't need HF accounts!
# Change these to your desired username/password
# For multiple users, use: auth=[("user1", "pass1"), ("user2", "pass2")]
demo.launch(
server_name="0.0.0.0",
server_port=7860,
auth=("starflow", "im30"), # Change password!
share=False # Set to True if you want public Gradio link
)