Optimising hparams

rvc/train/utils.py

@@ -2,7 +2,7 @@
 import glob
 import json
 import torch
-import argparse
+import sys
 import numpy as np
 from scipy.io.wavfile import read
 from collections import OrderedDict
@@ -245,113 +245,45 @@ def get_hparams():
     """
     Parses command line arguments and loads hyperparameters from a configuration file.
     """
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--save_every_epoch",
-        type=str,
-        help="Frequency (in epochs) at which checkpoints are saved.",
-    )
-    parser.add_argument(
-        "--total_epoch",
-        type=str,
-        help="Total number of training epochs.",
-    )
-    parser.add_argument(
-        "--pretrainG",
-        type=str,
-        help="Path to the pretrained Generator model.",
-    )
-    parser.add_argument(
-        "--pretrainD",
-        type=str,
-        help="Path to the pretrained Discriminator model.",
-    )
-    parser.add_argument(
-        "--gpus",
-        type=str,
-        help="Hyphen-separated list of GPU device IDs to use (e.g., '0-1-2').",
-    )
-    parser.add_argument(
-        "--batch_size",
-        type=str,
-        help="Batch size for training.",
-    )
-    parser.add_argument(
-        "--experiment_dir",
-        type=str,
-        help="Directory to store experiment outputs.",
-    )
-    parser.add_argument(
-        "--sample_rate",
-        type=str,
-        help="Sample rate to use.",
-    )
-    parser.add_argument(
-        "--save_every_weights",
-        type=str,
-        help="Save the model weights in the weights directory when saving checkpoints.",
-    )
-    parser.add_argument(
-        "--version",
-        type=str,
-        help="Model version identifier.",
-    )
-    parser.add_argument(
-        "--pitch_guidance",
-        type=str,
-        help="Use pitch (f0) as one of the inputs to the model (True or False).",
-    )
-    parser.add_argument(
-        "--if_latest",
-        type=str,
-        help="Only save the latest Generator/Discriminator model files (True or False).",
-    )
-    parser.add_argument(
-        "--if_cache_data_in_gpu",
-        type=str,
-        help="Cache the dataset in GPU memory (True or False).",
-    )
-    parser.add_argument(
-        "--overtraining_detector",
-        type=str,
-        help="Detect overtraining (True or False).",
-    )
-    parser.add_argument(
-        "--overtraining_threshold",
-        type=str,
-        help="Threshold for overtraining detection.",
-    )
-    parser.add_argument(
-        "--sync_graph",
-        type=str,
-        help="Synchronize graph (True or False).",
-    )
-
-    args = parser.parse_args()
-    name = args.experiment_dir
-    experiment_dir = os.path.join("./logs", args.experiment_dir)
+    args = {}
+    for i in range(1, len(sys.argv), 2):
+        if i + 1 < len(sys.argv):
+            key = sys.argv[i].replace('--', '')
+            value = sys.argv[i + 1]
+            if value.isdigit():
+                args[key] = int(value)
+            elif value.lower() == 'true':
+                args[key] = True
+            elif value.lower() == 'false':
+                args[key] = False
+            else:
+                args[key] = value
+
+    experiment_dir = os.path.join("./logs", args["experiment_dir"])
     config_save_path = os.path.join(experiment_dir, "config.json")
     with open(config_save_path, "r") as f:
         config = json.load(f)
+
     hparams = HParams(**config)
     hparams.model_dir = hparams.experiment_dir = experiment_dir
-    hparams.save_every_epoch = int(args.save_every_epoch)
-    hparams.name = name
-    hparams.total_epoch = int(args.total_epoch)
-    hparams.pretrainG = args.pretrainG
-    hparams.pretrainD = args.pretrainD
-    hparams.version = args.version
-    hparams.gpus = args.gpus
-    hparams.batch_size = int(args.batch_size)
-    hparams.sample_rate = int(args.sample_rate)
-    hparams.pitch_guidance = args.pitch_guidance
-    hparams.if_latest = bool(args.if_latest)
-    hparams.save_every_weights = bool(args.save_every_weights)
-    hparams.if_cache_data_in_gpu = bool(args.if_cache_data_in_gpu)
+    hparams.save_every_epoch = args.get("save_every_epoch")
+    hparams.name = args.get("experiment_dir")
+    hparams.total_epoch = args.get("total_epoch")
+    hparams.pretrainG = args.get("pretrainG")
+    hparams.pretrainD = args.get("pretrainD")
+    hparams.version = args.get("version")
+    hparams.gpus = args.get("gpus")
+    hparams.batch_size = args.get("batch_size")
+    hparams.sample_rate = args.get("sample_rate")
+    hparams.pitch_guidance = args.get("pitch_guidance")
+    hparams.if_latest = args.get("if_latest")
+    hparams.save_every_weights = args.get("save_every_weights")
+    hparams.if_cache_data_in_gpu = args.get("if_cache_data_in_gpu")
     hparams.data.training_files = f"{experiment_dir}/filelist.txt"
-    hparams.overtraining_detector = bool(args.overtraining_detector)
-    hparams.overtraining_threshold = int(args.overtraining_threshold)
-    hparams.sync_graph = args.sync_graph
+    hparams.overtraining_detector = args.get("overtraining_detector")
+    hparams.overtraining_threshold = args.get("overtraining_threshold")
+    hparams.sync_graph = args.get("sync_graph")
+
     print(hparams)
     return hparams