Fixed bug with train config

CHANGELOG.md

@@ -4,6 +4,8 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
+### Fixed
+- Bugs with using the train config functionality
 
 ## [0.1.4] - 2024-07-07
 ### Added
@@ -34,4 +36,3 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 [0.1.2]: https://github.com/adnaniazi/capfinder/compare/0.1.1...0.1.2
 [0.1.1]: https://pypi.org/manage/project/capfinder/release/0.1.1/
 [0.1.0]: https://pypi.org/manage/project/capfinder/release/0.1.0/
-

README.md

@@ -21,17 +21,17 @@ A package for decoding RNA cap types
 
 ## Installation
 
-# CPU installation
+### CPU installation
 ```sh
 pip install capfinder[cpu]
 ```
 
-# GPU installation (CUDA 12)
+### GPU installation (CUDA 12)
 ```sh
 pip install capfinder[gpu] "jax[cuda12]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
 ```
 
-# TPU installation
+### TPU installation
 ```sh
 pip install capfinder[tpu] "jax[tpu]" -f https://storage.googleapis.com/jax-releases/libtpu_releases.html
 ```
@@ -46,18 +46,18 @@ pip install capfinder[tpu] "jax[tpu]" -f https://storage.googleapis.com/jax-rele
   * Python 3.7+
 * Create a virtual environment and install the dependencies
 
-# CPU installation
+### CPU installation
 ```sh
 poetry install --extras cpu
 ```
 
-# GPU installation (CUDA 12)
+### GPU installation (CUDA 12)
 ```sh
 poetry install --extras gpu
 poetry run pip install "jax[cuda12]" -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
 ```
 
-# TPU installation
+### TPU installation
 ```sh
 poetry install --extras tpu
 poetry run pip install "jax[tpu]" -f https://storage.googleapis.com/jax-releases/libtpu_releases.html

pyproject.toml

@@ -18,7 +18,9 @@ classifiers = [
   "Operating System :: OS Independent",
   "Programming Language :: Python",
   "Programming Language :: Python :: 3",
-  "Programming Language :: Python :: 3.9",
+  "Programming Language :: Python :: 3.10",
+  "Programming Language :: Python :: 3.11",
+  "Programming Language :: Python :: 3.12",
   "Topic :: Software Development :: Libraries :: Python Modules",
   "Typing :: Typed",
 ]

src/capfinder/cli.py

@@ -255,8 +255,8 @@ def create_train_config(
             "dtype": "float16",
             "n_workers": 10,
             "n_classes": 4,
-            "use_local_dataset": True,
-            "remote_dataset_version": "latest",
+            "use_local_dataset": False,
+            "remote_dataset_version": "8.0.0",
         },
         "tune_params": {
             "comet_project_name": "capfinder_tfr_tune",

src/capfinder/training.py

@@ -235,6 +235,7 @@ def initialize_tuner(
     hyper_model: "CNNLSTMModel | EncoderModel",
     tune_params: dict,
     model_save_dir: str,
+    model_type: ModelType,
 ) -> Union[Hyperband, BayesianOptimization, RandomSearch]:
     """Initialize a Keras Tuner object based on the specified tuning strategy.
 
@@ -247,6 +248,8 @@ def initialize_tuner(
     model_save_dir: str
         The directory where the model should be saved.
     comet_project_name: str
+    model_type: ModelType
+        Type of the model to be trained.
 
     Returns:
     --------
@@ -465,16 +468,14 @@ def run_training_pipeline(
         input_shape=(etl_params["target_length"], 1), n_classes=etl_params["n_classes"]
     )
 
-    tuner = initialize_tuner(hyper_model, tune_params, model_save_dir)
+    tuner = initialize_tuner(hyper_model, tune_params, model_save_dir, model_type)
 
-    tensorboard_save_path = os.path.join(model_save_dir, "tensorboard_logs_encoder")
-    logger.info(
-        f"Run tensorboard as following:\ntensorboard --logdir {tensorboard_save_path}"
-    )
+    # tensorboard_save_path = os.path.join(model_save_dir, "tensorboard_logs_encoder", model_type)
+    # logger.info(
+    #     f"Run tensorboard as following:\ntensorboard --logdir {tensorboard_save_path}"
+    # )
 
     # Split train into train-val sets
-    # dataset_size = len(list(train_dataset))  # Calculate the dataset size
-
     dataset_size = train_dataset.reduce(0, lambda x, _: x + 1).numpy()
 
     train_size = int(0.8 * dataset_size)  # 80% for training
@@ -494,7 +495,6 @@ def run_training_pipeline(
             validation_data=val_dataset,
             validation_steps=val_size,
             epochs=tune_params["max_epochs_hpt"],
-            # batch_size=tune_params["batch_size"],
             callbacks=[
                 keras.callbacks.EarlyStopping(
                     patience=tune_params["patience"], restore_best_weights=True
@@ -722,10 +722,10 @@ def run_training_pipeline(
             text=conf_matrix_str_train,
             metadata={"Description": "Train Confusion Matrix"},
         )
-        train_experiment.log_text(
-            text=f"tensorboard --logdir {tensorboard_save_path}",
-            metadata={"Description": "Tensorboard command"},
-        )
+        # train_experiment.log_text(
+        #     text=f"tensorboard --logdir {tensorboard_save_path}",
+        #     metadata={"Description": "Tensorboard command"},
+        # )
 
         # Log the test set confusion matrix to the Comet ML dashboard pane
         train_experiment.log_confusion_matrix(