podstack 1.2.0__py3-none-any.whl

podstack/registry/model_utils.py

@@ -0,0 +1,231 @@
+"""
+Model Utilities
+
+Framework-agnostic model serialization and deserialization helpers.
+All framework imports are lazy — no hard dependencies on any ML framework.
+"""
+
+import os
+import pickle
+from typing import Any, Optional
+
+from .exceptions import ModelSerializationError, FrameworkNotInstalledError
+
+
+# Framework detection mapping: module prefix -> framework name
+_FRAMEWORK_MODULE_MAP = {
+    "torch": "pytorch",
+    "tensorflow": "tensorflow",
+    "keras": "tensorflow",
+    "sklearn": "sklearn",
+    "transformers": "huggingface",
+}
+
+
+def detect_framework(model: Any) -> str:
+    """
+    Auto-detect the ML framework from a model object.
+
+    Args:
+        model: A model object from any supported framework.
+
+    Returns:
+        Framework name string: "pytorch", "tensorflow", "sklearn", "huggingface", or "pickle".
+    """
+    module = type(model).__module__ or ""
+
+    for prefix, framework in _FRAMEWORK_MODULE_MAP.items():
+        if module.startswith(prefix):
+            return framework
+
+    # Fallback: check for common base classes by name to avoid imports
+    type_name = type(model).__name__
+    type_mro = [cls.__module__ + "." + cls.__name__ for cls in type(model).__mro__]
+
+    for entry in type_mro:
+        for prefix, framework in _FRAMEWORK_MODULE_MAP.items():
+            if entry.startswith(prefix):
+                return framework
+
+    return "pickle"
+
+
+def save_model(model: Any, path: str, framework: Optional[str] = None) -> str:
+    """
+    Serialize a model to disk.
+
+    Args:
+        model: The model object to save.
+        path: Directory path to save the model into.
+        framework: Framework name. Auto-detected if not provided.
+            One of: "pytorch", "tensorflow", "sklearn", "huggingface", "pickle".
+
+    Returns:
+        Path to the saved model file/directory.
+
+    Raises:
+        FrameworkNotInstalledError: If the required framework is not installed.
+        ModelSerializationError: If serialization fails.
+    """
+    if framework is None:
+        framework = detect_framework(model)
+
+    os.makedirs(path, exist_ok=True)
+
+    try:
+        if framework == "pytorch":
+            return _save_pytorch(model, path)
+        elif framework == "tensorflow":
+            return _save_tensorflow(model, path)
+        elif framework == "sklearn":
+            return _save_sklearn(model, path)
+        elif framework == "huggingface":
+            return _save_huggingface(model, path)
+        else:
+            return _save_pickle(model, path)
+    except FrameworkNotInstalledError:
+        raise
+    except Exception as e:
+        raise ModelSerializationError(f"Failed to save {framework} model: {e}")
+
+
+def load_model_from_path(path: str, framework: str) -> Any:
+    """
+    Deserialize a model from disk.
+
+    Args:
+        path: Path to the saved model file or directory.
+        framework: Framework name used to save the model.
+            One of: "pytorch", "tensorflow", "sklearn", "huggingface", "pickle".
+
+    Returns:
+        The loaded model object.
+
+    Raises:
+        FrameworkNotInstalledError: If the required framework is not installed.
+        ModelSerializationError: If deserialization fails.
+    """
+    try:
+        if framework == "pytorch":
+            return _load_pytorch(path)
+        elif framework == "tensorflow":
+            return _load_tensorflow(path)
+        elif framework == "sklearn":
+            return _load_sklearn(path)
+        elif framework == "huggingface":
+            return _load_huggingface(path)
+        else:
+            return _load_pickle(path)
+    except FrameworkNotInstalledError:
+        raise
+    except Exception as e:
+        raise ModelSerializationError(f"Failed to load {framework} model: {e}")
+
+
+# ==================== PyTorch ====================
+
+def _save_pytorch(model: Any, path: str) -> str:
+    try:
+        import torch
+    except ImportError:
+        raise FrameworkNotInstalledError("torch")
+
+    filepath = os.path.join(path, "model.pt")
+    torch.save(model.state_dict() if hasattr(model, "state_dict") else model, filepath)
+    return filepath
+
+
+def _load_pytorch(path: str) -> Any:
+    try:
+        import torch
+    except ImportError:
+        raise FrameworkNotInstalledError("torch")
+
+    filepath = path if os.path.isfile(path) else os.path.join(path, "model.pt")
+    return torch.load(filepath, map_location="cpu", weights_only=False)
+
+
+# ==================== TensorFlow / Keras ====================
+
+def _save_tensorflow(model: Any, path: str) -> str:
+    try:
+        import tensorflow  # noqa: F401
+    except ImportError:
+        raise FrameworkNotInstalledError("tensorflow")
+
+    save_dir = os.path.join(path, "saved_model")
+    model.save(save_dir)
+    return save_dir
+
+
+def _load_tensorflow(path: str) -> Any:
+    try:
+        import tensorflow as tf
+    except ImportError:
+        raise FrameworkNotInstalledError("tensorflow")
+
+    load_dir = path if os.path.isdir(path) and os.path.exists(os.path.join(path, "saved_model.pb")) else os.path.join(path, "saved_model")
+    return tf.keras.models.load_model(load_dir)
+
+
+# ==================== Scikit-learn ====================
+
+def _save_sklearn(model: Any, path: str) -> str:
+    try:
+        import sklearn  # noqa: F401
+    except ImportError:
+        raise FrameworkNotInstalledError("sklearn")
+
+    filepath = os.path.join(path, "model.pkl")
+    with open(filepath, "wb") as f:
+        pickle.dump(model, f)
+    return filepath
+
+
+def _load_sklearn(path: str) -> Any:
+    try:
+        import sklearn  # noqa: F401
+    except ImportError:
+        raise FrameworkNotInstalledError("sklearn")
+
+    filepath = path if os.path.isfile(path) else os.path.join(path, "model.pkl")
+    with open(filepath, "rb") as f:
+        return pickle.load(f)
+
+
+# ==================== Hugging Face ====================
+
+def _save_huggingface(model: Any, path: str) -> str:
+    try:
+        import transformers  # noqa: F401
+    except ImportError:
+        raise FrameworkNotInstalledError("huggingface")
+
+    save_dir = os.path.join(path, "hf_model")
+    model.save_pretrained(save_dir)
+    return save_dir
+
+
+def _load_huggingface(path: str) -> Any:
+    try:
+        from transformers import AutoModel
+    except ImportError:
+        raise FrameworkNotInstalledError("huggingface")
+
+    load_dir = path if os.path.isdir(path) and os.path.exists(os.path.join(path, "config.json")) else os.path.join(path, "hf_model")
+    return AutoModel.from_pretrained(load_dir)
+
+
+# ==================== Pickle (fallback) ====================
+
+def _save_pickle(model: Any, path: str) -> str:
+    filepath = os.path.join(path, "model.pkl")
+    with open(filepath, "wb") as f:
+        pickle.dump(model, f)
+    return filepath
+
+
+def _load_pickle(path: str) -> Any:
+    filepath = path if os.path.isfile(path) else os.path.join(path, "model.pkl")
+    with open(filepath, "rb") as f:
+        return pickle.load(f)

podstack-1.2.0.dist-info/METADATA

@@ -0,0 +1,299 @@
+Metadata-Version: 2.4
+Name: podstack
+Version: 1.2.0
+Summary: Official Python SDK for Podstack GPU Notebook Platform
+Author-email: Podstack <support@podstack.io>
+License-Expression: MIT
+Project-URL: Homepage, https://podstack.io
+Project-URL: Documentation, https://docs.podstack.io
+Project-URL: Repository, https://github.com/podstack/podstack-python
+Project-URL: Issues, https://github.com/podstack/podstack-python/issues
+Keywords: gpu,notebook,machine-learning,deep-learning,cloud,jupyter
+Classifier: Development Status :: 5 - Production/Stable
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: Science/Research
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.8
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Libraries :: Python Modules
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: httpx>=0.24.0
+Requires-Dist: requests>=2.28.0
+Provides-Extra: torch
+Requires-Dist: torch; extra == "torch"
+Provides-Extra: tensorflow
+Requires-Dist: tensorflow; extra == "tensorflow"
+Provides-Extra: sklearn
+Requires-Dist: scikit-learn; extra == "sklearn"
+Provides-Extra: huggingface
+Requires-Dist: transformers; extra == "huggingface"
+Requires-Dist: safetensors; extra == "huggingface"
+Provides-Extra: all
+Requires-Dist: torch; extra == "all"
+Requires-Dist: tensorflow; extra == "all"
+Requires-Dist: scikit-learn; extra == "all"
+Requires-Dist: transformers; extra == "all"
+Requires-Dist: safetensors; extra == "all"
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0.0; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.21.0; extra == "dev"
+Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
+Requires-Dist: black>=23.0.0; extra == "dev"
+Requires-Dist: mypy>=1.0.0; extra == "dev"
+Requires-Dist: ruff>=0.0.270; extra == "dev"
+Dynamic: license-file
+
+# Podstack Python SDK
+
+Official Python SDK for the Podstack GPU Notebook Platform. Launch GPU notebooks in under 1 second and execute ML workloads with ease.
+
+## Installation
+
+```bash
+pip install podstack
+```
+
+## Quick Start
+
+```python
+import asyncio
+from podstack import Client
+
+async def main():
+    async with Client(api_key="your-api-key") as client:
+        # Create a GPU notebook
+        notebook = await client.notebooks.create(
+            name="my-experiment",
+            gpu_type="A100",
+            environment="pytorch"
+        )
+
+        # Execute code
+        result = await notebook.execute("""
+import torch
+print(f"GPU: {torch.cuda.get_device_name(0)}")
+print(f"Memory: {torch.cuda.get_device_properties(0).total_memory / 1e9:.1f} GB")
+        """)
+
+        print(result.output)
+
+        # Save a version
+        version = await notebook.save(message="Initial experiment")
+
+        # Stop when done
+        await notebook.stop()
+
+asyncio.run(main())
+```
+
+## Sync Usage
+
+For simple scripts, use the sync wrappers:
+
+```python
+from podstack import Client
+
+client = Client(api_key="your-api-key")
+
+# Create notebook
+notebook = client.sync_create_notebook(name="quick-test", gpu_type="A10")
+
+# Run code
+result = client.sync_run("print('Hello GPU!')", gpu_type="A10")
+print(result.output)
+```
+
+## Features
+
+### Notebooks
+
+```python
+# Create with options
+notebook = await client.notebooks.create(
+    name="training-run",
+    gpu_type="A100",
+    environment="pytorch",
+    project_id="proj_xxx",
+    idle_timeout_minutes=60,
+    auto_shutdown_enabled=True,
+    metadata={"experiment": "v2"}
+)
+
+# List notebooks
+notebooks = await client.notebooks.list(status="running")
+
+# Execute code
+result = await notebook.execute("import torch; print(torch.cuda.device_count())")
+
+# Access JupyterLab
+print(f"JupyterLab: {notebook.jupyter_url}")
+
+# Stop/Start
+await notebook.stop()
+await notebook.start()
+```
+
+### Serverless Executions
+
+Run code without managing notebooks:
+
+```python
+# Quick execution
+result = await client.executions.run(
+    code="print('Hello!')",
+    gpu_type="A10",
+    environment="pytorch"
+)
+
+# Non-blocking execution
+execution = await client.executions.create(
+    code=long_running_code,
+    gpu_type="H100",
+    timeout_seconds=3600
+)
+
+# Check status later
+await execution.refresh()
+if execution.is_complete:
+    print(execution.output)
+```
+
+### Notebook Versioning
+
+Git-like versioning for notebooks:
+
+```python
+# Save a version
+version = await notebook.save(message="Added training loop")
+
+# List versions
+versions = await notebook.list_versions()
+
+# Restore a version
+await notebook.restore_version(version.id)
+
+# Create a branch
+await notebook.create_branch("experiment-v2", from_version_id=version.id)
+```
+
+### Projects
+
+Organize notebooks into projects:
+
+```python
+# Create project
+project = await client.create_project(
+    name="ML Research",
+    description="Transformer experiments"
+)
+
+# Create notebook in project
+notebook = await client.notebooks.create(
+    name="attention-study",
+    gpu_type="A100",
+    project_id=project.id
+)
+
+# List project notebooks
+notebooks = await client.notebooks.list(project_id=project.id)
+```
+
+### Billing & Usage
+
+```python
+# Check balance
+balance = await client.get_wallet_balance()
+print(f"Balance: ₹{balance.balance:.2f}")
+
+# Get usage
+usage = await client.get_usage(
+    start_date="2024-01-01",
+    end_date="2024-01-31",
+    group_by="day"
+)
+print(f"Total cost: ₹{usage.total_cost:.2f}")
+```
+
+### GPU Types
+
+```python
+# List available GPUs
+gpus = await client.list_gpus()
+for gpu in gpus:
+    print(f"{gpu.type}: {gpu.memory_gb}GB, ₹{gpu.price_per_hour_paise/100:.2f}/hr")
+```
+
+Available GPU types:
+- `T4` - 16GB, budget-friendly
+- `L4` - 24GB, inference optimized
+- `A10` - 24GB, balanced
+- `A100_40GB` - 40GB, training
+- `A100_80GB` - 80GB, large models
+- `H100` - 80GB, fastest
+
+### Webhooks
+
+```python
+# Create webhook
+webhook = await client.create_webhook(
+    url="https://your-server.com/webhook",
+    events=["notebook.started", "execution.completed"]
+)
+
+# List webhooks
+webhooks = await client.list_webhooks()
+```
+
+## Error Handling
+
+```python
+from podstack import (
+    Client,
+    PodstackError,
+    AuthenticationError,
+    GPUNotAvailableError,
+    RateLimitError,
+    ExecutionTimeoutError
+)
+
+try:
+    async with Client(api_key="invalid") as client:
+        await client.notebooks.create(name="test", gpu_type="A100")
+except AuthenticationError:
+    print("Invalid API key")
+except GPUNotAvailableError as e:
+    print(f"GPU {e.gpu_type} not available, try: {e.available_types}")
+except RateLimitError as e:
+    print(f"Rate limited, retry after {e.retry_after}s")
+except ExecutionTimeoutError as e:
+    print(f"Execution {e.execution_id} timed out")
+except PodstackError as e:
+    print(f"Error [{e.code}]: {e.message}")
+```
+
+## Configuration
+
+```python
+# Environment variables
+# PODSTACK_API_KEY=psk_live_xxxxx
+# PODSTACK_BASE_URL=https://api.podstack.io/v1
+
+# Or pass directly
+client = Client(
+    api_key="psk_live_xxxxx",
+    base_url="https://api.podstack.io/v1",
+    timeout=60.0,
+    max_retries=5
+)
+```
+
+## License
+
+MIT License - see LICENSE for details.

podstack-1.2.0.dist-info/RECORD

@@ -0,0 +1,27 @@
+podstack/__init__.py,sha256=WLwTzeSi8QFkhGjFeXS85vg9KnDlyOxefYhwRfd9mZE,5026
+podstack/annotations.py,sha256=7O68bjF-djeNNZbJxG0-gWNxuTTx8CL-Sqqp5Mpo6sw,23308
+podstack/client.py,sha256=8_Zl7TX19XGvM_BDXwI-U10GgxLqYvvCYERUA526nB0,10838
+podstack/exceptions.py,sha256=02e-9BjYlO0t2wwCJ9XFWqhX-VbaH31gfmUpoqr2KH8,4288
+podstack/execution.py,sha256=CHtRw_xWQzYgZJOfGL40CRtduZQnppkmFNw0NmTKsu8,8948
+podstack/gpu_runner.py,sha256=BDvdrVNyTh7oeOqpvUuq2xSj53xD4TU-bdo-h88JO_o,41047
+podstack/models.py,sha256=qMD5kIyJ30eysZhgcVR4XgQzOFina1ZtFJ8qwr8ThxE,7552
+podstack/notebook.py,sha256=Q2ZrdZ1wYaAKyrRLcCZW3EPV2K8V76_OxDpzLeFItv4,12425
+podstack/registry/__init__.py,sha256=HfzDP6rIPSvEfgoBckPjAj7D8TYHb_NHmZwJnnKOUuY,9989
+podstack/registry/client.py,sha256=Q1vI4scH1CgeQZb_lYcaGEuMzdFn1p8P6thuqMyLd7c,31537
+podstack/registry/exceptions.py,sha256=NyRufm3nA_8wypIMqo6k_oo2re0L5cddIZWsAxftYgw,3408
+podstack/registry/experiment.py,sha256=V_n1Er37jghXamUq9wtZg7KMpdY2XvuCEH7BklwLhXo,7375
+podstack/registry/model.py,sha256=JT7mZf2kKiguzp-S0e5fC1UlLjvGe7_-QrH074yCAkI,9673
+podstack/registry/model_utils.py,sha256=XD5ckSXUd-sVLAAFlSGx66l8X_2WJO31D_sRQQKaG1I,6872
+podstack-1.2.0.dist-info/licenses/LICENSE,sha256=NlFavTiYs5FYzcXNNBZFBob8ih5Ktj79h1oyCAHjkD8,1070
+podstack_gpu/__init__.py,sha256=uIRZ2MGniy2G-UoJzKLxdCC-3d3ef01h5osrCWPLC-k,3674
+podstack_gpu/app.py,sha256=eMTIOBPyv01JCwngtflZhE_OuYn2i9Ox7gmCQ7HA-pI,21357
+podstack_gpu/exceptions.py,sha256=oOWM0cdShby5ZQnlGP2xTxl8lgU9rdS4WKBw5OKvs38,757
+podstack_gpu/image.py,sha256=4D8apxNM66PoBpzdk488B6zb03heWRgwR0g9o3f5vyM,10387
+podstack_gpu/runner.py,sha256=jk78MMi0DCHyuDeTuBQvQwTW93GzQiUH0sTxG0691WI,25565
+podstack_gpu/secret.py,sha256=kKFbfVrPETmkbhpJwAZPeIQ-btCFd-XX9gEtkMyoSjM,5331
+podstack_gpu/utils.py,sha256=vAa0S_zUHOs4cjs3Qsbk-o5Sji9VA7mfYGNjI8JikKc,6639
+podstack_gpu/volume.py,sha256=N_t8qlW94l7rh8Uq2uvJKEvhGobzEY7Bcw5emdyLauQ,5533
+podstack-1.2.0.dist-info/METADATA,sha256=nNthxAu_tLzIEUQzwQn5NlH89QZhVLPzKhZw2z6oeSg,7334
+podstack-1.2.0.dist-info/WHEEL,sha256=YCfwYGOYMi5Jhw2fU4yNgwErybb2IX5PEwBKV4ZbdBo,91
+podstack-1.2.0.dist-info/top_level.txt,sha256=Mj_82W07f_sLiodgYSGpa86Lke95K7RP0uE5aNE4qzY,22
+podstack-1.2.0.dist-info/RECORD,,

podstack-1.2.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (82.0.0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

podstack-1.2.0.dist-info/licenses/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024-2026 Podstack
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

podstack-1.2.0.dist-info/top_level.txt

@@ -0,0 +1,2 @@
+podstack
+podstack_gpu