added app.yaml configs

.riahub/workflows/workflow.yaml

@@ -82,13 +82,13 @@ jobs:
       - name: 4. Convert to ONNX file
         run: |
           mkdir -p onnx_files
-          MKL_DISABLE_FAST_MM=1 PYTHONPATH=. python onnx_scripts/convert_to_onnx.py
+          MKL_DISABLE_FAST_MM=1 PYTHONPATH=. python scripts/onnx/convert_to_onnx.py
           echo "building inference app"
           
       - name: Upload ONNX file
         uses: actions/upload-artifact@v3
         with:
-          name: ria-demo-onnx
+          name: onnx-file
           path: onnx_files/inference_recognition_model.onnx
 
       - name: List checkpoint directory
@@ -97,7 +97,7 @@ jobs:
 
       - name: 5. Profile ONNX model
         run: |
-          PYTHONPATH=. python onnx_scripts/profile_onnx.py
+          PYTHONPATH=. python scrips/onnx/profile_onnx.py
       
       - name: Upload JSON profiling data
         uses: actions/upload-artifact@v3
@@ -107,16 +107,12 @@ jobs:
       
       - name: 6. Convert to ORT file
         run: |
-          python -m onnxruntime.tools.convert_onnx_models_to_ort \
-             /workspace/qoherent/modrec-workflow/onnx_files/inference_recognition_model.onnx \
-            --output_dir ort_files \
-            --optimization_style Fixed \
-            --target_platform amd64
+          python -m scripts/ort/convert_to_ort.py
 
 
       - name: Upload ORT file
         uses: actions/upload-artifact@v3
         with:
-          name: ria-demo-ort
+          name: ort-file
           path: ort_files/inference_recognition_model.ort
           

conf/app.yaml

@@ -16,25 +16,52 @@ dataset:
   #multiple modulations to contain in the dataset
   modulation_types: [bpsk, qpsk, qam16, qam64]
 
+  # Rolloff factor for pulse shaping filter (0 < beta <= 1)
+  beta: 0.3
+
+  # Samples per symbol (determines bandwidth of the digital signal)
+  sps: 4
+
+  # SNR sweep range: start, stop (exclusive), and step (in dB)
+  snr_start: -6     # Start value of SNR sweep (in dB)
+  snr_stop: 13      # Stop value (exclusive) of SNR sweep (in dB)
+  snr_step: 3       # Step size for SNR sweep (in dB)
+
+  # Number of iterations (samples) per modulation and SNR combination
+  num_iterations: 3
+
+  # Number of samples per generated recording
+  recording_length: 1024
+
+
 
 training:
-  #number of training samples being processed together before model updates its weights
+  # Number of training samples processed together before the model updates its weights
   batch_size: 256
-  
-  #number of passes through the data set during the training process
+
+  # Number of complete passes through the training dataset during training
   epochs: 5
 
-  #how much the weights update during training after every batch
-  #suggested range for fine-tuning: (1e-6, 1e-4)
+  # Learning rate: how much weights are updated after every batch
+  # Suggested range for fine-tuning: 1e-6 to 1e-4
   learning_rate: 1e-4
 
+  # Whether to use GPU acceleration for training (if available)
   use_gpu: true
 
-inference:
-  #num classes to classify on
-  num_classes: 4
+  # Dropout rate for individual neurons/layers (probability of dropping out a unit)
+  drop_rate: 0.5
 
+  # Drop path rate: probability of dropping entire residual paths (stochastic depth)
+  drop_path_rate: 0.2
+
+  # Weight decay (L2 regularization) to help prevent overfitting
+  wd: 0.01
 
 
 app:
-  build_dir: dist
+  # Optimization style for ORT conversion. Options: 'Fixed', 'None'
+  optimization_style: Fixed
+
+  # Target platform architecture. Common options: 'amd64', 'arm64'
+  target_platform: amd64

helpers/app_settings.py

@@ -17,6 +17,13 @@ class DataSetConfig:
     seed: int
     modulation_types: list
     val_split: float
+    beta: float
+    sps: int
+    snr_start: int
+    snr_end: int
+    snr_step: int
+    num_iterations: int
+    recording_length: int
 
 
 @dataclass
@@ -25,16 +32,16 @@ class TrainingConfig:
     epochs: int
     learning_rate: float
     use_gpu: bool
-
-
-@dataclass
-class InferenceConfig:
-    num_classes: int
+    drop_rate: float
+    drop_path_rate: float
+    wd: int
+    
 
 
 @dataclass
 class AppConfig:
-    build_dir: str
+    optimization_style: str
+    target_platform: str
 
 
 class AppSettings:
@@ -49,7 +56,6 @@ class AppSettings:
         self.general = GeneralConfig(**config_data["general"])
         self.dataset = DataSetConfig(**config_data["dataset"])
         self.training = TrainingConfig(**config_data["training"])
-        self.inference = InferenceConfig(**config_data["inference"])
         self.app = AppConfig(**config_data["app"])
 
 

requirements.txt

@@ -1,12 +1,46 @@
-h5py
-pytorch-lightning
-matplotlib
-numpy
-PyYAML
-scikit-learn
-timm
-torch
-onnx
-onnxruntime
+aiohappyeyeballs~=2.6.1
+aiohttp~=3.12.12
+aiosignal~=1.3.2
+attrs~=25.3.0
+filelock~=3.18.0
+frozenlist~=1.7.0
+fsspec~=2025.5.1
+h5py~=3.14.0
+idna~=3.10
+Jinja2~=3.1.6
+lightning~=2.5.1.post0
+lightning-utilities~=0.14.3
+MarkupSafe~=3.0.2
+mpmath~=1.3.0
+multidict~=6.4.4
+networkx~=3.5
+numpy~=2.3.0
+nvidia-cublas-cu12~=12.6.4.1
+nvidia-cuda-cupti-cu12~=12.6.80
+nvidia-cuda-nvrtc-cu12~=12.6.77
+nvidia-cuda-runtime-cu12~=12.6.77
+nvidia-cudnn-cu12~=9.5.1.17
+nvidia-cufft-cu12~=11.3.0.4
+nvidia-cufile-cu12~=1.11.1.6
+nvidia-curand-cu12~=10.3.7.77
+nvidia-cusolver-cu12~=11.7.1.2
+nvidia-cusparse-cu12~=12.5.4.2
+nvidia-cusparselt-cu12~=0.6.3
+nvidia-nccl-cu12~=2.26.2
+nvidia-nvjitlink-cu12~=12.6.85
+nvidia-nvtx-cu12~=12.6.77
+packaging~=24.2
+propcache~=0.3.2
+pytorch-lightning~=2.5.1.post0
+PyYAML~=6.0.2
+setuptools~=80.9.0
+sympy~=1.14.0
+torch~=2.7.1
+torchmetrics~=1.7.3
+tqdm~=4.67.1
+triton~=3.3.1
+typing_extensions~=4.14.0
+utils~=1.0.2
+yarl~=1.20.1
 --index-url https://git.riahub.ai/api/packages/qoherent/pypi/simple/ 
 utils

scripts/dataset_building/data_gen.py

@@ -3,6 +3,7 @@ import numpy as np
 from utils.signal import block_generator
 import argparse
 import os
+from helpers.app_settings import get_app_settings
 
 mods = {
     "bpsk": {"num_bits_per_symbol": 1, "constellation_type": "psk"},
@@ -13,12 +14,13 @@ mods = {
 
 
 def generate_modulated_signals(output_dir):
-    for modulation in ["bpsk", "qpsk", "qam16", "qam64"]:
-        for snr in np.arange(-6, 13, 3):
+    settings = get_app_settings().dataset
+    for modulation in settings.modulation_types:
+        for snr in np.arange(settings.snr_start, settings.snr_end, settings.snr_step):
             for i in range(3):
-                recording_length = 1024
-                beta = 0.3  # the rolloff factor, can be changed to add variety
-                sps = 4  # samples per symbol, or the relative bandwidth of the digital signal. Can also be changed.
+                recording_length = settings.recording_length
+                beta = settings.beta  # the rolloff factor, can be changed to add variety
+                sps = settings.sps  # samples per symbol, or the relative bandwidth of the digital signal. Can also be changed.
 
                 # blocks don't directly take the string 'qpsk' so we use the dict 'mods' to get parameters
                 constellation_type = mods[modulation]["constellation_type"]

scripts/onnx/convert_to_onnx.py

@@ -17,14 +17,13 @@ def convert_to_onnx(ckpt_path, fp16=False):
         output_path (str): The path to save the converted ONNX model.
     """
     settings = get_app_settings()
-
-    inference_cfg = settings.inference
+    
     dataset_cfg = settings.dataset
 
     in_channels = 2
     batch_size = 1
     slice_length = int(1024 / dataset_cfg.num_slices)
-    num_classes = inference_cfg.num_classes
+    num_classes = len(dataset_cfg.modulation_types)
 
     model = RFClassifier(
         model=mobilenetv3(

scripts/ort/convert_to_ort.py

@@ -0,0 +1,24 @@
+import subprocess
+from helpers.app_settings import get_app_settings
+
+settings = get_app_settings()
+
+input_path = f"{settings.app.build_dir}/inference_recognition_model.onnx"
+
+
+optimization_style = settings.app.optimization_style
+target_platform = settings.app.optimization_style
+
+# Build the command
+command = [
+    "python",
+    "-m",
+    "onnxruntime.tools.convert_onnx_models_to_ort",
+    input_path,
+    "--output_dir", "ort_files",
+    "--optimization_style", optimization_style,
+    "--target_platform", target_platform
+]
+
+# Run it
+subprocess.run(command, check=True)

scripts/training/train.py

@@ -60,10 +60,10 @@ def train_model():
     num_classes = len(ds_train.label_encoder.classes_)
 
     hparams = {
-        "drop_path_rate": 0.2,
-        "drop_rate": 0.5,
+        "drop_path_rate": training_cfg.drop_path_rate,
+        "drop_rate": training_cfg.drop_rate,
         "learning_rate": float(training_cfg.learning_rate),
-        "wd": 0.01,
+        "wd": training_cfg.wd,
     }
 
     class RFClassifier(L.LightningModule):