Commit
·
31ad4fe
1
Parent(s):
ab2fedd
Refactor atlas-export.py for improved readability and maintainability
Browse files- atlas-export.py +93 -75
atlas-export.py
CHANGED
|
@@ -65,6 +65,7 @@ def check_gpu_available() -> bool:
|
|
| 65 |
"""Check if GPU is available for computation."""
|
| 66 |
try:
|
| 67 |
import torch
|
|
|
|
| 68 |
return torch.cuda.is_available()
|
| 69 |
except ImportError:
|
| 70 |
return False
|
|
@@ -74,52 +75,58 @@ def build_atlas_command(args) -> list:
|
|
| 74 |
"""Build the embedding-atlas command with all parameters."""
|
| 75 |
# Use uvx to run embedding-atlas with required dependencies
|
| 76 |
# Include hf-transfer for faster downloads when HF_HUB_ENABLE_HF_TRANSFER is set
|
| 77 |
-
cmd = [
|
| 78 |
-
|
| 79 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
# Add all optional parameters
|
| 81 |
if args.model:
|
| 82 |
cmd.extend(["--model", args.model])
|
| 83 |
-
|
| 84 |
# Always specify text column to avoid interactive prompt
|
| 85 |
text_col = args.text_column or "text" # Default to "text" if not specified
|
| 86 |
cmd.extend(["--text", text_col])
|
| 87 |
-
|
| 88 |
if args.image_column:
|
| 89 |
cmd.extend(["--image", args.image_column])
|
| 90 |
-
|
| 91 |
if args.split:
|
| 92 |
cmd.extend(["--split", args.split])
|
| 93 |
-
|
| 94 |
if args.sample:
|
| 95 |
cmd.extend(["--sample", str(args.sample)])
|
| 96 |
-
|
| 97 |
if args.trust_remote_code:
|
| 98 |
cmd.append("--trust-remote-code")
|
| 99 |
-
|
| 100 |
if not args.compute_embeddings:
|
| 101 |
cmd.append("--no-compute-embeddings")
|
| 102 |
-
|
| 103 |
if args.x_column:
|
| 104 |
cmd.extend(["--x", args.x_column])
|
| 105 |
-
|
| 106 |
if args.y_column:
|
| 107 |
cmd.extend(["--y", args.y_column])
|
| 108 |
-
|
| 109 |
if args.neighbors_column:
|
| 110 |
cmd.extend(["--neighbors", args.neighbors_column])
|
| 111 |
-
|
| 112 |
# Add export flag with output path
|
| 113 |
export_path = "atlas_export.zip"
|
| 114 |
cmd.extend(["--export-application", export_path])
|
| 115 |
-
|
| 116 |
return cmd, export_path
|
| 117 |
|
| 118 |
|
| 119 |
def create_space_readme(args) -> str:
|
| 120 |
"""Generate README.md content for the Space."""
|
| 121 |
title = args.space_name.replace("-", " ").title()
|
| 122 |
-
|
| 123 |
readme = f"""---
|
| 124 |
title: {title}
|
| 125 |
emoji: 🗺️
|
|
@@ -141,13 +148,13 @@ Interactive embedding visualization of [{args.dataset_id}](https://huggingface.c
|
|
| 141 |
- Automatic clustering with labels
|
| 142 |
- WebGPU-accelerated rendering
|
| 143 |
"""
|
| 144 |
-
|
| 145 |
if args.model:
|
| 146 |
readme += f"\n## Model\n\nEmbeddings generated using: `{args.model}`\n"
|
| 147 |
-
|
| 148 |
if args.sample:
|
| 149 |
readme += f"\n## Data\n\nVisualization includes {args.sample:,} samples from the dataset.\n"
|
| 150 |
-
|
| 151 |
readme += """
|
| 152 |
## How to Use
|
| 153 |
|
|
@@ -161,21 +168,21 @@ Interactive embedding visualization of [{args.dataset_id}](https://huggingface.c
|
|
| 161 |
|
| 162 |
*Generated with [UV Scripts Atlas Export](https://huggingface.co/uv-scripts)*
|
| 163 |
"""
|
| 164 |
-
|
| 165 |
return readme
|
| 166 |
|
| 167 |
|
| 168 |
def extract_and_prepare_static_files(zip_path: str, output_dir: Path) -> None:
|
| 169 |
"""Extract the exported atlas ZIP and prepare for static deployment."""
|
| 170 |
logger.info(f"Extracting {zip_path} to {output_dir}")
|
| 171 |
-
|
| 172 |
-
with zipfile.ZipFile(zip_path,
|
| 173 |
zip_ref.extractall(output_dir)
|
| 174 |
-
|
| 175 |
# The ZIP should contain index.html and associated files
|
| 176 |
if not (output_dir / "index.html").exists():
|
| 177 |
raise FileNotFoundError("index.html not found in exported atlas")
|
| 178 |
-
|
| 179 |
logger.info(f"Extracted {len(list(output_dir.iterdir()))} items")
|
| 180 |
|
| 181 |
|
|
@@ -184,11 +191,11 @@ def deploy_to_space(
|
|
| 184 |
space_name: str,
|
| 185 |
organization: Optional[str] = None,
|
| 186 |
private: bool = False,
|
| 187 |
-
hf_token: Optional[str] = None
|
| 188 |
) -> str:
|
| 189 |
"""Deploy the static files to a HuggingFace Space."""
|
| 190 |
api = HfApi(token=hf_token)
|
| 191 |
-
|
| 192 |
# Construct full repo ID
|
| 193 |
if organization:
|
| 194 |
repo_id = f"{organization}/{space_name}"
|
|
@@ -197,9 +204,9 @@ def deploy_to_space(
|
|
| 197 |
user_info = api.whoami()
|
| 198 |
username = user_info["name"]
|
| 199 |
repo_id = f"{username}/{space_name}"
|
| 200 |
-
|
| 201 |
logger.info(f"Creating Space: {repo_id}")
|
| 202 |
-
|
| 203 |
# Create the Space repository
|
| 204 |
try:
|
| 205 |
create_repo(
|
|
@@ -207,7 +214,7 @@ def deploy_to_space(
|
|
| 207 |
repo_type="space",
|
| 208 |
space_sdk="static",
|
| 209 |
private=private,
|
| 210 |
-
token=hf_token
|
| 211 |
)
|
| 212 |
logger.info(f"Created new Space: {repo_id}")
|
| 213 |
except Exception as e:
|
|
@@ -215,39 +222,36 @@ def deploy_to_space(
|
|
| 215 |
logger.info(f"Space {repo_id} already exists, updating...")
|
| 216 |
else:
|
| 217 |
raise
|
| 218 |
-
|
| 219 |
# Upload all files
|
| 220 |
logger.info("Uploading files to Space...")
|
| 221 |
upload_folder(
|
| 222 |
-
folder_path=str(output_dir),
|
| 223 |
-
repo_id=repo_id,
|
| 224 |
-
repo_type="space",
|
| 225 |
-
token=hf_token
|
| 226 |
)
|
| 227 |
-
|
| 228 |
space_url = f"https://huggingface.co/spaces/{repo_id}"
|
| 229 |
logger.info(f"✅ Space deployed successfully: {space_url}")
|
| 230 |
-
|
| 231 |
return space_url
|
| 232 |
|
| 233 |
|
| 234 |
def main():
|
| 235 |
# Enable HF Transfer for faster downloads if available
|
| 236 |
os.environ.setdefault("HF_HUB_ENABLE_HF_TRANSFER", "1")
|
| 237 |
-
|
| 238 |
parser = argparse.ArgumentParser(
|
| 239 |
description="Generate and deploy static Embedding Atlas visualizations",
|
| 240 |
formatter_class=argparse.RawDescriptionHelpFormatter,
|
| 241 |
epilog=__doc__,
|
| 242 |
)
|
| 243 |
-
|
| 244 |
# Required arguments
|
| 245 |
parser.add_argument(
|
| 246 |
"dataset_id",
|
| 247 |
type=str,
|
| 248 |
help="HuggingFace dataset ID to visualize",
|
| 249 |
)
|
| 250 |
-
|
| 251 |
# Space configuration
|
| 252 |
parser.add_argument(
|
| 253 |
"--space-name",
|
|
@@ -265,7 +269,7 @@ def main():
|
|
| 265 |
action="store_true",
|
| 266 |
help="Make the Space private",
|
| 267 |
)
|
| 268 |
-
|
| 269 |
# Atlas configuration
|
| 270 |
parser.add_argument(
|
| 271 |
"--model",
|
|
@@ -298,7 +302,7 @@ def main():
|
|
| 298 |
action="store_true",
|
| 299 |
help="Trust remote code in dataset/model",
|
| 300 |
)
|
| 301 |
-
|
| 302 |
# Pre-computed embeddings
|
| 303 |
parser.add_argument(
|
| 304 |
"--no-compute-embeddings",
|
|
@@ -321,7 +325,7 @@ def main():
|
|
| 321 |
type=str,
|
| 322 |
help="Column with neighbor indices (for pre-computed)",
|
| 323 |
)
|
| 324 |
-
|
| 325 |
# Additional options
|
| 326 |
parser.add_argument(
|
| 327 |
"--hf-token",
|
|
@@ -338,44 +342,52 @@ def main():
|
|
| 338 |
type=str,
|
| 339 |
help="Local directory for output (default: temp directory)",
|
| 340 |
)
|
| 341 |
-
|
| 342 |
args = parser.parse_args()
|
| 343 |
-
|
| 344 |
# Check GPU availability
|
| 345 |
if check_gpu_available():
|
| 346 |
logger.info("🚀 GPU detected - may accelerate embedding generation")
|
| 347 |
else:
|
| 348 |
logger.info("💻 Running on CPU - embedding generation may be slower")
|
| 349 |
-
|
| 350 |
# Login to HuggingFace if needed
|
| 351 |
if not args.local_only:
|
| 352 |
# Try to get token from various sources
|
| 353 |
hf_token = (
|
| 354 |
-
args.hf_token
|
| 355 |
or os.environ.get("HF_TOKEN")
|
| 356 |
or os.environ.get("HUGGING_FACE_HUB_TOKEN")
|
| 357 |
or get_token() # This will check HF CLI login
|
| 358 |
)
|
| 359 |
-
|
| 360 |
if hf_token:
|
| 361 |
login(token=hf_token)
|
| 362 |
logger.info("✅ Authenticated with Hugging Face")
|
| 363 |
else:
|
| 364 |
# Check if running in non-interactive environment (HF Jobs, CI, etc.)
|
| 365 |
is_interactive = sys.stdin.isatty()
|
| 366 |
-
|
| 367 |
if is_interactive:
|
| 368 |
-
logger.warning(
|
|
|
|
|
|
|
| 369 |
response = input("Continue anyway? (y/n): ")
|
| 370 |
-
if response.lower() !=
|
| 371 |
sys.exit(0)
|
| 372 |
else:
|
| 373 |
# In non-interactive environments, fail immediately if no token
|
| 374 |
-
logger.error(
|
| 375 |
-
|
| 376 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 377 |
sys.exit(1)
|
| 378 |
-
|
| 379 |
# Set up output directory
|
| 380 |
if args.output_dir:
|
| 381 |
output_dir = Path(args.output_dir)
|
|
@@ -385,50 +397,48 @@ def main():
|
|
| 385 |
temp_dir = tempfile.mkdtemp(prefix="atlas_export_")
|
| 386 |
output_dir = Path(temp_dir)
|
| 387 |
logger.info(f"Using temporary directory: {output_dir}")
|
| 388 |
-
|
| 389 |
try:
|
| 390 |
# Build and run embedding-atlas command
|
| 391 |
cmd, export_path = build_atlas_command(args)
|
| 392 |
logger.info(f"Running command: {' '.join(cmd)}")
|
| 393 |
-
|
| 394 |
# Run the command
|
| 395 |
result = subprocess.run(cmd, capture_output=True, text=True)
|
| 396 |
-
|
| 397 |
if result.returncode != 0:
|
| 398 |
logger.error(f"Atlas export failed with return code {result.returncode}")
|
| 399 |
logger.error(f"STDOUT: {result.stdout}")
|
| 400 |
logger.error(f"STDERR: {result.stderr}")
|
| 401 |
sys.exit(1)
|
| 402 |
-
|
| 403 |
logger.info("✅ Atlas export completed successfully")
|
| 404 |
-
|
| 405 |
# Extract the exported files
|
| 406 |
extract_and_prepare_static_files(export_path, output_dir)
|
| 407 |
-
|
| 408 |
# Create README for the Space
|
| 409 |
readme_content = create_space_readme(args)
|
| 410 |
(output_dir / "README.md").write_text(readme_content)
|
| 411 |
-
|
| 412 |
if args.local_only:
|
| 413 |
logger.info(f"✅ Static files prepared in: {output_dir}")
|
| 414 |
-
logger.info(
|
|
|
|
|
|
|
| 415 |
else:
|
| 416 |
# Deploy to HuggingFace Space
|
| 417 |
space_url = deploy_to_space(
|
| 418 |
-
output_dir,
|
| 419 |
-
args.space_name,
|
| 420 |
-
args.organization,
|
| 421 |
-
args.private,
|
| 422 |
-
hf_token
|
| 423 |
)
|
| 424 |
-
|
| 425 |
logger.info(f"\n🎉 Success! Your atlas is live at: {space_url}")
|
| 426 |
logger.info(f"The visualization will be available in a few moments.")
|
| 427 |
-
|
| 428 |
# Clean up the ZIP file
|
| 429 |
if Path(export_path).exists():
|
| 430 |
os.remove(export_path)
|
| 431 |
-
|
| 432 |
finally:
|
| 433 |
# Clean up temp directory if used
|
| 434 |
if temp_dir and not args.local_only:
|
|
@@ -443,12 +453,20 @@ if __name__ == "__main__":
|
|
| 443 |
print("# Basic usage:")
|
| 444 |
print("uv run atlas-export.py stanfordnlp/imdb --space-name imdb-atlas\n")
|
| 445 |
print("# With custom model and sampling:")
|
| 446 |
-
print(
|
|
|
|
|
|
|
| 447 |
print("# For HF Jobs with GPU (experimental UV support):")
|
| 448 |
-
print(
|
| 449 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 450 |
print("# Local generation only:")
|
| 451 |
-
print(
|
|
|
|
|
|
|
| 452 |
sys.exit(0)
|
| 453 |
-
|
| 454 |
-
main()
|
|
|
|
| 65 |
"""Check if GPU is available for computation."""
|
| 66 |
try:
|
| 67 |
import torch
|
| 68 |
+
|
| 69 |
return torch.cuda.is_available()
|
| 70 |
except ImportError:
|
| 71 |
return False
|
|
|
|
| 75 |
"""Build the embedding-atlas command with all parameters."""
|
| 76 |
# Use uvx to run embedding-atlas with required dependencies
|
| 77 |
# Include hf-transfer for faster downloads when HF_HUB_ENABLE_HF_TRANSFER is set
|
| 78 |
+
cmd = [
|
| 79 |
+
"uvx",
|
| 80 |
+
"--with",
|
| 81 |
+
"datasets",
|
| 82 |
+
"--with",
|
| 83 |
+
"hf-transfer",
|
| 84 |
+
"embedding-atlas",
|
| 85 |
+
args.dataset_id,
|
| 86 |
+
]
|
| 87 |
# Add all optional parameters
|
| 88 |
if args.model:
|
| 89 |
cmd.extend(["--model", args.model])
|
| 90 |
+
|
| 91 |
# Always specify text column to avoid interactive prompt
|
| 92 |
text_col = args.text_column or "text" # Default to "text" if not specified
|
| 93 |
cmd.extend(["--text", text_col])
|
| 94 |
+
|
| 95 |
if args.image_column:
|
| 96 |
cmd.extend(["--image", args.image_column])
|
| 97 |
+
|
| 98 |
if args.split:
|
| 99 |
cmd.extend(["--split", args.split])
|
| 100 |
+
|
| 101 |
if args.sample:
|
| 102 |
cmd.extend(["--sample", str(args.sample)])
|
| 103 |
+
|
| 104 |
if args.trust_remote_code:
|
| 105 |
cmd.append("--trust-remote-code")
|
| 106 |
+
|
| 107 |
if not args.compute_embeddings:
|
| 108 |
cmd.append("--no-compute-embeddings")
|
| 109 |
+
|
| 110 |
if args.x_column:
|
| 111 |
cmd.extend(["--x", args.x_column])
|
| 112 |
+
|
| 113 |
if args.y_column:
|
| 114 |
cmd.extend(["--y", args.y_column])
|
| 115 |
+
|
| 116 |
if args.neighbors_column:
|
| 117 |
cmd.extend(["--neighbors", args.neighbors_column])
|
| 118 |
+
|
| 119 |
# Add export flag with output path
|
| 120 |
export_path = "atlas_export.zip"
|
| 121 |
cmd.extend(["--export-application", export_path])
|
| 122 |
+
|
| 123 |
return cmd, export_path
|
| 124 |
|
| 125 |
|
| 126 |
def create_space_readme(args) -> str:
|
| 127 |
"""Generate README.md content for the Space."""
|
| 128 |
title = args.space_name.replace("-", " ").title()
|
| 129 |
+
|
| 130 |
readme = f"""---
|
| 131 |
title: {title}
|
| 132 |
emoji: 🗺️
|
|
|
|
| 148 |
- Automatic clustering with labels
|
| 149 |
- WebGPU-accelerated rendering
|
| 150 |
"""
|
| 151 |
+
|
| 152 |
if args.model:
|
| 153 |
readme += f"\n## Model\n\nEmbeddings generated using: `{args.model}`\n"
|
| 154 |
+
|
| 155 |
if args.sample:
|
| 156 |
readme += f"\n## Data\n\nVisualization includes {args.sample:,} samples from the dataset.\n"
|
| 157 |
+
|
| 158 |
readme += """
|
| 159 |
## How to Use
|
| 160 |
|
|
|
|
| 168 |
|
| 169 |
*Generated with [UV Scripts Atlas Export](https://huggingface.co/uv-scripts)*
|
| 170 |
"""
|
| 171 |
+
|
| 172 |
return readme
|
| 173 |
|
| 174 |
|
| 175 |
def extract_and_prepare_static_files(zip_path: str, output_dir: Path) -> None:
|
| 176 |
"""Extract the exported atlas ZIP and prepare for static deployment."""
|
| 177 |
logger.info(f"Extracting {zip_path} to {output_dir}")
|
| 178 |
+
|
| 179 |
+
with zipfile.ZipFile(zip_path, "r") as zip_ref:
|
| 180 |
zip_ref.extractall(output_dir)
|
| 181 |
+
|
| 182 |
# The ZIP should contain index.html and associated files
|
| 183 |
if not (output_dir / "index.html").exists():
|
| 184 |
raise FileNotFoundError("index.html not found in exported atlas")
|
| 185 |
+
|
| 186 |
logger.info(f"Extracted {len(list(output_dir.iterdir()))} items")
|
| 187 |
|
| 188 |
|
|
|
|
| 191 |
space_name: str,
|
| 192 |
organization: Optional[str] = None,
|
| 193 |
private: bool = False,
|
| 194 |
+
hf_token: Optional[str] = None,
|
| 195 |
) -> str:
|
| 196 |
"""Deploy the static files to a HuggingFace Space."""
|
| 197 |
api = HfApi(token=hf_token)
|
| 198 |
+
|
| 199 |
# Construct full repo ID
|
| 200 |
if organization:
|
| 201 |
repo_id = f"{organization}/{space_name}"
|
|
|
|
| 204 |
user_info = api.whoami()
|
| 205 |
username = user_info["name"]
|
| 206 |
repo_id = f"{username}/{space_name}"
|
| 207 |
+
|
| 208 |
logger.info(f"Creating Space: {repo_id}")
|
| 209 |
+
|
| 210 |
# Create the Space repository
|
| 211 |
try:
|
| 212 |
create_repo(
|
|
|
|
| 214 |
repo_type="space",
|
| 215 |
space_sdk="static",
|
| 216 |
private=private,
|
| 217 |
+
token=hf_token,
|
| 218 |
)
|
| 219 |
logger.info(f"Created new Space: {repo_id}")
|
| 220 |
except Exception as e:
|
|
|
|
| 222 |
logger.info(f"Space {repo_id} already exists, updating...")
|
| 223 |
else:
|
| 224 |
raise
|
| 225 |
+
|
| 226 |
# Upload all files
|
| 227 |
logger.info("Uploading files to Space...")
|
| 228 |
upload_folder(
|
| 229 |
+
folder_path=str(output_dir), repo_id=repo_id, repo_type="space", token=hf_token
|
|
|
|
|
|
|
|
|
|
| 230 |
)
|
| 231 |
+
|
| 232 |
space_url = f"https://huggingface.co/spaces/{repo_id}"
|
| 233 |
logger.info(f"✅ Space deployed successfully: {space_url}")
|
| 234 |
+
|
| 235 |
return space_url
|
| 236 |
|
| 237 |
|
| 238 |
def main():
|
| 239 |
# Enable HF Transfer for faster downloads if available
|
| 240 |
os.environ.setdefault("HF_HUB_ENABLE_HF_TRANSFER", "1")
|
| 241 |
+
|
| 242 |
parser = argparse.ArgumentParser(
|
| 243 |
description="Generate and deploy static Embedding Atlas visualizations",
|
| 244 |
formatter_class=argparse.RawDescriptionHelpFormatter,
|
| 245 |
epilog=__doc__,
|
| 246 |
)
|
| 247 |
+
|
| 248 |
# Required arguments
|
| 249 |
parser.add_argument(
|
| 250 |
"dataset_id",
|
| 251 |
type=str,
|
| 252 |
help="HuggingFace dataset ID to visualize",
|
| 253 |
)
|
| 254 |
+
|
| 255 |
# Space configuration
|
| 256 |
parser.add_argument(
|
| 257 |
"--space-name",
|
|
|
|
| 269 |
action="store_true",
|
| 270 |
help="Make the Space private",
|
| 271 |
)
|
| 272 |
+
|
| 273 |
# Atlas configuration
|
| 274 |
parser.add_argument(
|
| 275 |
"--model",
|
|
|
|
| 302 |
action="store_true",
|
| 303 |
help="Trust remote code in dataset/model",
|
| 304 |
)
|
| 305 |
+
|
| 306 |
# Pre-computed embeddings
|
| 307 |
parser.add_argument(
|
| 308 |
"--no-compute-embeddings",
|
|
|
|
| 325 |
type=str,
|
| 326 |
help="Column with neighbor indices (for pre-computed)",
|
| 327 |
)
|
| 328 |
+
|
| 329 |
# Additional options
|
| 330 |
parser.add_argument(
|
| 331 |
"--hf-token",
|
|
|
|
| 342 |
type=str,
|
| 343 |
help="Local directory for output (default: temp directory)",
|
| 344 |
)
|
| 345 |
+
|
| 346 |
args = parser.parse_args()
|
| 347 |
+
|
| 348 |
# Check GPU availability
|
| 349 |
if check_gpu_available():
|
| 350 |
logger.info("🚀 GPU detected - may accelerate embedding generation")
|
| 351 |
else:
|
| 352 |
logger.info("💻 Running on CPU - embedding generation may be slower")
|
| 353 |
+
|
| 354 |
# Login to HuggingFace if needed
|
| 355 |
if not args.local_only:
|
| 356 |
# Try to get token from various sources
|
| 357 |
hf_token = (
|
| 358 |
+
args.hf_token
|
| 359 |
or os.environ.get("HF_TOKEN")
|
| 360 |
or os.environ.get("HUGGING_FACE_HUB_TOKEN")
|
| 361 |
or get_token() # This will check HF CLI login
|
| 362 |
)
|
| 363 |
+
|
| 364 |
if hf_token:
|
| 365 |
login(token=hf_token)
|
| 366 |
logger.info("✅ Authenticated with Hugging Face")
|
| 367 |
else:
|
| 368 |
# Check if running in non-interactive environment (HF Jobs, CI, etc.)
|
| 369 |
is_interactive = sys.stdin.isatty()
|
| 370 |
+
|
| 371 |
if is_interactive:
|
| 372 |
+
logger.warning(
|
| 373 |
+
"No HF token provided. You may not be able to push to the Hub."
|
| 374 |
+
)
|
| 375 |
response = input("Continue anyway? (y/n): ")
|
| 376 |
+
if response.lower() != "y":
|
| 377 |
sys.exit(0)
|
| 378 |
else:
|
| 379 |
# In non-interactive environments, fail immediately if no token
|
| 380 |
+
logger.error(
|
| 381 |
+
"No HF token found. Cannot deploy to Space in non-interactive environment."
|
| 382 |
+
)
|
| 383 |
+
logger.error(
|
| 384 |
+
"Please set HF_TOKEN environment variable or use --hf-token argument."
|
| 385 |
+
)
|
| 386 |
+
logger.error(
|
| 387 |
+
"Checked: HF_TOKEN, HUGGING_FACE_HUB_TOKEN, and HF CLI login"
|
| 388 |
+
)
|
| 389 |
sys.exit(1)
|
| 390 |
+
|
| 391 |
# Set up output directory
|
| 392 |
if args.output_dir:
|
| 393 |
output_dir = Path(args.output_dir)
|
|
|
|
| 397 |
temp_dir = tempfile.mkdtemp(prefix="atlas_export_")
|
| 398 |
output_dir = Path(temp_dir)
|
| 399 |
logger.info(f"Using temporary directory: {output_dir}")
|
| 400 |
+
|
| 401 |
try:
|
| 402 |
# Build and run embedding-atlas command
|
| 403 |
cmd, export_path = build_atlas_command(args)
|
| 404 |
logger.info(f"Running command: {' '.join(cmd)}")
|
| 405 |
+
|
| 406 |
# Run the command
|
| 407 |
result = subprocess.run(cmd, capture_output=True, text=True)
|
| 408 |
+
|
| 409 |
if result.returncode != 0:
|
| 410 |
logger.error(f"Atlas export failed with return code {result.returncode}")
|
| 411 |
logger.error(f"STDOUT: {result.stdout}")
|
| 412 |
logger.error(f"STDERR: {result.stderr}")
|
| 413 |
sys.exit(1)
|
| 414 |
+
|
| 415 |
logger.info("✅ Atlas export completed successfully")
|
| 416 |
+
|
| 417 |
# Extract the exported files
|
| 418 |
extract_and_prepare_static_files(export_path, output_dir)
|
| 419 |
+
|
| 420 |
# Create README for the Space
|
| 421 |
readme_content = create_space_readme(args)
|
| 422 |
(output_dir / "README.md").write_text(readme_content)
|
| 423 |
+
|
| 424 |
if args.local_only:
|
| 425 |
logger.info(f"✅ Static files prepared in: {output_dir}")
|
| 426 |
+
logger.info(
|
| 427 |
+
"To deploy manually, upload the contents to a HuggingFace Space with sdk: static"
|
| 428 |
+
)
|
| 429 |
else:
|
| 430 |
# Deploy to HuggingFace Space
|
| 431 |
space_url = deploy_to_space(
|
| 432 |
+
output_dir, args.space_name, args.organization, args.private, hf_token
|
|
|
|
|
|
|
|
|
|
|
|
|
| 433 |
)
|
| 434 |
+
|
| 435 |
logger.info(f"\n🎉 Success! Your atlas is live at: {space_url}")
|
| 436 |
logger.info(f"The visualization will be available in a few moments.")
|
| 437 |
+
|
| 438 |
# Clean up the ZIP file
|
| 439 |
if Path(export_path).exists():
|
| 440 |
os.remove(export_path)
|
| 441 |
+
|
| 442 |
finally:
|
| 443 |
# Clean up temp directory if used
|
| 444 |
if temp_dir and not args.local_only:
|
|
|
|
| 453 |
print("# Basic usage:")
|
| 454 |
print("uv run atlas-export.py stanfordnlp/imdb --space-name imdb-atlas\n")
|
| 455 |
print("# With custom model and sampling:")
|
| 456 |
+
print(
|
| 457 |
+
"uv run atlas-export.py my-dataset --space-name my-viz --model nomic-ai/nomic-embed-text-v1.5 --sample 10000\n"
|
| 458 |
+
)
|
| 459 |
print("# For HF Jobs with GPU (experimental UV support):")
|
| 460 |
+
print(
|
| 461 |
+
'# First get your token: python -c "from huggingface_hub import get_token; print(get_token())"'
|
| 462 |
+
)
|
| 463 |
+
print(
|
| 464 |
+
"hf jobs uv run --flavor t4-small -s HF_TOKEN=your-token-here https://huggingface.co/datasets/uv-scripts/build-atlas/raw/main/atlas-export.py dataset --space-name viz --model sentence-transformers/all-mpnet-base-v2\n"
|
| 465 |
+
)
|
| 466 |
print("# Local generation only:")
|
| 467 |
+
print(
|
| 468 |
+
"uv run atlas-export.py dataset --space-name test --local-only --output-dir ./atlas-output"
|
| 469 |
+
)
|
| 470 |
sys.exit(0)
|
| 471 |
+
|
| 472 |
+
main()
|