[ISSUE 534]: Deprecate conda environment.yml / move to pip

docs/source/installation.rst

@@ -3,46 +3,76 @@
 Local Installation
 ==================
 
-Step 1: Get Miniconda
-^^^^^^^^^^^^^^^^^^^^^
+Step 1: Install Python
+^^^^^^^^^^^^^^^^^^^^^^
 
-This is a Python package, and it depends on other packages to work. To manage 
-all this, it's recommended to use Miniconda. Get it from 
-https://docs.anaconda.com/miniconda/
+Install Python 3.9 or newer.
 
-Step 2: Install NAM
-^^^^^^^^^^^^^^^^^^^
+Step 2: Create a virtual environment
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+It's recommended to install NAM in a virtual environment.
+
+On macOS / Linux:
+
+.. code-block:: console
+
+   $ python -m venv .venv
+   $ source .venv/bin/activate
+   $ python -m pip install --upgrade pip
+
+On Windows, in ``cmd.exe``:
+
+.. code-block:: console
 
-Now that we have Miniconda, we can install NAM using it.
+   > python -m venv .venv
+   > .venv\Scripts\activate
+   > python -m pip install --upgrade pip
 
-(Windows / Linux users) If your computer has an nVIDIA GPU, you should install a
-GPU-compatible version of PyTorch first. 
-`The PyTorch website <https://pytorch.org/get-started/locally/>`_ will always
-have the most up-to-date guidance for this. Currently, this is the command:
+On Windows, in PowerShell:
 
 .. code-block:: console
 
-   $ pip install torch --index-url https://download.pytorch.org/whl/cu129
+   > python -m venv .venv
+   > .\.venv\Scripts\Activate.ps1
+   > python -m pip install --upgrade pip
+
+Step 3: Install PyTorch
+^^^^^^^^^^^^^^^^^^^^^^^
+
+If your computer has an NVIDIA GPU, install a GPU-compatible version of
+PyTorch using the instructions on the PyTorch website:
 
-Then, install NAM using pip:
+https://pytorch.org/get-started/locally/
+
+Step 4: Install NAM
+^^^^^^^^^^^^^^^^^^^
 
 .. code-block:: console
 
-   $ pip install neural-amp-modeler
+   $ python -m pip install neural-amp-modeler
 
 To update an existing installation:
 
 .. code-block:: console
 
-   pip install --upgrade neural-amp-modeler
+   $ python -m pip install --upgrade neural-amp-modeler
 
 Local development installation
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-If you're interested in developing this package, there are Anaconda environment
-definitions included in the ``environments/`` directory. Use the one that's
-appropriate for the platform you're developing on. The
-``.github/workflows/python-pckage.yml`` is also helpful if you want to be sure
+Create and activate a virtual environment, then install NAM in editable mode
+with its test dependencies. On Windows, use the activation command above for
+your shell. Install any additional development tooling you need alongside it:
+
+.. code-block:: console
+
+   $ python -m venv .venv
+   $ source .venv/bin/activate
+   $ python -m pip install --upgrade pip
+   $ python -m pip install -e ".[dev,test]"
+
+``.github/workflows/python-package.yml`` is also helpful if you want to be sure
 that you're testing your developments in the same way that contributions will be
 automatically tested (via GitHub Actions).
 
@@ -51,7 +81,7 @@ Trouble using the GPU?
 ^^^^^^^^^^^^^^^^^^^^^^
 
 If you're using a Windows or Linux machine with an NVIDIA GPU and NAM isn't
-using it (Apple machines with Apple Silicon don't use an nVIDIA GPU, but MPS, an 
+using it (Apple machines with Apple Silicon don't use an NVIDIA GPU, but MPS, an 
 accelerator with somewhat similar functionality), the reason is 99.999% probably
 an issue with your PyTorch installation, not NAM. Google (or ChatGPT) should be 
 able to help you fix the issue, but here are a few handy things you can do (in 
@@ -81,7 +111,7 @@ To uninstall PyTorch and reinstall it, you can do:
 
 .. code-block:: console
 
-   $ pip uninstall torch torchvision torchaudio
+   $ python -m pip uninstall torch torchvision torchaudio
 
 and then use the install command above (or check the PyTorch website for the
 most up-to-date instructions). If you notice that this documentation is out of 

pyproject.toml

@@ -36,6 +36,12 @@ transformers-compat = [
     "numpy",  # NB: Older PyTorch versions require numpy<2
 ]
 
+dev = [
+    "black",
+    "flake8",
+    "pre-commit",
+]
+
 test = [
     "pytest",
     "pytest-mock",

tests/test_graceful_shutdown.py

@@ -238,7 +238,6 @@ def run_training_with_interrupt(
     interrupt_after_training_starts: bool = True,
     interrupt_delay: float = 2.0,
     timeout: float = 120.0,
-    conda_env: Optional[str] = None,
 ) -> Tuple[int, bool, list]:
     """
     Run nam-full training and send SIGINT after training starts.
@@ -248,36 +247,17 @@ def run_training_with_interrupt(
     :param interrupt_delay: Seconds to wait after training starts (or after
         process start if interrupt_after_training_starts is False)
     :param timeout: Maximum time to wait for process to complete
-    :param conda_env: Name of conda environment to use
     :return: (exit_code, graceful_shutdown_detected, nam_files_found)
     """
-    # Build command
-    if conda_env:
-        # Use conda run to execute in the environment
-        cmd = [
-            "conda",
-            "run",
-            "-n",
-            conda_env,
-            "--no-capture-output",
-            "nam-full",
-            str(data_config_path),
-            str(model_config_path),
-            str(learning_config_path),
-            str(output_path),
-            "--no-show",
-            "--no-plots",
-        ]
-    else:
-        cmd = [
-            "nam-full",
-            str(data_config_path),
-            str(model_config_path),
-            str(learning_config_path),
-            str(output_path),
-            "--no-show",
-            "--no-plots",
-        ]
+    cmd = [
+        "nam-full",
+        str(data_config_path),
+        str(model_config_path),
+        str(learning_config_path),
+        str(output_path),
+        "--no-show",
+        "--no-plots",
+    ]
 
     print(f"Starting training with command: {' '.join(cmd)}")
 
@@ -387,7 +367,7 @@ def run_training_with_interrupt(
     return exit_code, graceful_shutdown_detected, nam_files
 
 
-def test_graceful_shutdown(conda_env: Optional[str] = None) -> bool:
+def test_graceful_shutdown() -> bool:
     """
     Test that graceful shutdown generates a model file.
 
@@ -422,7 +402,6 @@ def test_graceful_shutdown(conda_env: Optional[str] = None) -> bool:
             interrupt_after_training_starts=True,
             interrupt_delay=2.0,  # Wait 2s after training starts
             timeout=120.0,
-            conda_env=conda_env,
         )
 
         print("\n" + "=" * 60)
@@ -491,7 +470,6 @@ def test_graceful_shutdown_generates_model(self):
                 interrupt_after_training_starts=True,
                 interrupt_delay=2.0,  # Wait 2s after training starts
                 timeout=120.0,
-                conda_env=None,  # Use current environment in pytest
             )
 
             # Assert that a .nam file was created
@@ -506,18 +484,11 @@ def main():
     """Run the graceful shutdown test."""
     import argparse
 
-    parser = argparse.ArgumentParser(
+    argparse.ArgumentParser(
         description="Test graceful shutdown model generation"
-    )
-    parser.add_argument(
-        "--conda-env",
-        type=str,
-        default=None,
-        help="Conda environment to use for running nam-full",
-    )
-    args = parser.parse_args()
+    ).parse_args()
 
-    success = test_graceful_shutdown(conda_env=args.conda_env)
+    success = test_graceful_shutdown()
     sys.exit(0 if success else 1)