halib 0.1.67__py3-none-any.whl → 0.1.73__py3-none-any.whl

halib/research/base_config.py

@@ -0,0 +1,100 @@
+import os
+from rich.pretty import pprint
+from abc import ABC, abstractmethod
+from dataclass_wizard import YAMLWizard
+
+
+class NamedConfig(ABC):
+    """
+    Base class for named configurations.
+    All configurations should have a name.
+    """
+
+    @abstractmethod
+    def get_name(self):
+        """
+        Get the name of the configuration.
+        This method should be implemented in subclasses.
+        """
+        pass
+
+
+class ExpBaseConfig(ABC, YAMLWizard):
+    """
+    Base class for configuration objects.
+    What a cfg class must have:
+    1 - a dataset cfg
+    2 - a metric cfg
+    3 - a method cfg
+    """
+
+    # Save to yaml fil
+    def save_to_outdir(
+        self, filename: str = "config.yaml", outdir=None, override: bool = False
+    ) -> None:
+        """
+        Save the configuration to the output directory.
+        """
+        if outdir is not None:
+            output_dir = outdir
+        else:
+            output_dir = self.get_outdir()
+        os.makedirs(output_dir, exist_ok=True)
+        assert (output_dir is not None) and (
+            os.path.isdir(output_dir)
+        ), f"Output directory '{output_dir}' does not exist or is not a directory."
+        file_path = os.path.join(output_dir, filename)
+        if os.path.exists(file_path) and not override:
+            pprint(
+                f"File '{file_path}' already exists. Use 'override=True' to overwrite."
+            )
+        else:
+            # method of YAMLWizard to_yaml_file
+            self.to_yaml_file(file_path)
+
+    @classmethod
+    @abstractmethod
+    # load from a custom YAML file
+    def from_custom_yaml_file(cls, yaml_file: str):
+        """Load a configuration from a custom YAML file."""
+        pass
+
+    @abstractmethod
+    def get_cfg_name(self):
+        """
+        Get the name of the configuration.
+        This method should be implemented in subclasses.
+        """
+        pass
+
+    @abstractmethod
+    def get_outdir(self):
+        """
+        Get the output directory for the configuration.
+        This method should be implemented in subclasses.
+        """
+        return None
+
+    @abstractmethod
+    def get_general_cfg(self):
+        """
+        Get the general configuration like output directory, log settings, SEED, etc.
+        This method should be implemented in subclasses.
+        """
+        pass
+
+    @abstractmethod
+    def get_dataset_cfg(self) -> NamedConfig:
+        """
+        Get the dataset configuration.
+        This method should be implemented in subclasses.
+        """
+        pass
+
+    @abstractmethod
+    def get_metric_cfg(self) -> NamedConfig:
+        """
+        Get the metric configuration.
+        This method should be implemented in subclasses.
+        """
+        pass

halib/research/base_exp.py

@@ -0,0 +1,99 @@
+from abc import ABC, abstractmethod
+
+from ..research.base_config import ExpBaseConfig
+from ..research.perfcalc import PerfCalc
+from ..research.metrics import MetricsBackend
+
+# ! SEE https://github.com/hahv/base_exp for sample usage
+class BaseExperiment(PerfCalc, ABC):
+    """
+    Base class for experiments.
+    Orchestrates the experiment pipeline using a pluggable metrics backend.
+    """
+
+    def __init__(self, config: ExpBaseConfig):
+        self.config = config
+        self.metric_backend = None
+
+    # -----------------------
+    # PerfCalc Required Methods
+    # -----------------------
+    def get_dataset_name(self):
+        return self.config.get_dataset_cfg().get_name()
+
+    def get_experiment_name(self):
+        return self.config.get_cfg_name()
+
+    def get_metric_backend(self):
+        if not self.metric_backend:
+            self.metric_backend = self.prepare_metrics(self.config.get_metric_cfg())
+        return self.metric_backend
+
+    # -----------------------
+    # Abstract Experiment Steps
+    # -----------------------
+    @abstractmethod
+    def init_general(self, general_cfg):
+        """Setup general settings like SEED, logging, env variables."""
+        pass
+
+    @abstractmethod
+    def prepare_dataset(self, dataset_cfg):
+        """Load/prepare dataset."""
+        pass
+
+    @abstractmethod
+    def prepare_metrics(self, metric_cfg) -> MetricsBackend:
+        """
+        Prepare the metrics for the experiment.
+        This method should be implemented in subclasses.
+        """
+        pass
+
+    @abstractmethod
+    def exec_exp(self, *args, **kwargs):
+        """Run experiment process, e.g.: training/evaluation loop.
+        Return: raw_metrics_data, and extra_data as input for calc_and_save_exp_perfs
+        """
+        pass
+
+    def eval_exp(self):
+        """Optional: re-run evaluation from saved results."""
+        pass
+
+    # -----------------------
+    # Main Experiment Runner
+    # -----------------------
+    def run_exp(self, do_calc_metrics=True, *args, **kwargs):
+        """
+        Run the whole experiment pipeline.
+        Params:
+            + 'outfile' to save csv file results,
+            + 'outdir' to set output directory for experiment results.
+            + 'return_df' to return a DataFrame of results instead of a dictionary.
+
+        Full pipeline:
+            1. Init
+            2. Dataset
+            3. Metrics Preparation
+            4. Save Config
+            5. Execute
+            6. Calculate & Save Metrics
+        """
+        self.init_general(self.config.get_general_cfg())
+        self.prepare_dataset(self.config.get_dataset_cfg())
+
+        # Save config before running
+        self.config.save_to_outdir()
+
+        # Execute experiment
+        results = self.exec_exp(*args, **kwargs)
+        if do_calc_metrics:
+            metrics_data, extra_data = results
+            # Calculate & Save metrics
+            perf_results = self.calc_and_save_exp_perfs(
+                raw_metrics_data=metrics_data, extra_data=extra_data, *args, **kwargs
+            )
+            return perf_results
+        else:
+            return results

{halib-0.1.67.dist-info → halib-0.1.73.dist-info}/METADATA

@@ -1,190 +1,189 @@
-Metadata-Version: 2.4
-Name: halib
-Version: 0.1.67
-Summary: Small library for common tasks
-Author: Hoang Van Ha
-Author-email: hoangvanhauit@gmail.com
-Classifier: Programming Language :: Python :: 3
-Classifier: License :: OSI Approved :: MIT License
-Classifier: Operating System :: OS Independent
-Requires-Python: >=3.8
-Description-Content-Type: text/markdown
-License-File: LICENSE.txt
-Requires-Dist: arrow
-Requires-Dist: click
-Requires-Dist: enlighten
-Requires-Dist: kaleido==0.1.*
-Requires-Dist: loguru
-Requires-Dist: more-itertools
-Requires-Dist: moviepy
-Requires-Dist: networkx
-Requires-Dist: numpy
-Requires-Dist: omegaconf
-Requires-Dist: opencv-python
-Requires-Dist: pandas
-Requires-Dist: Pillow
-Requires-Dist: Pyarrow
-Requires-Dist: pycurl
-Requires-Dist: python-telegram-bot
-Requires-Dist: requests
-Requires-Dist: rich
-Requires-Dist: scikit-learn
-Requires-Dist: matplotlib
-Requires-Dist: seaborn
-Requires-Dist: plotly
-Requires-Dist: pygwalker
-Requires-Dist: tabulate
-Requires-Dist: itables
-Requires-Dist: timebudget
-Requires-Dist: tqdm
-Requires-Dist: tube_dl
-Requires-Dist: wandb
-Requires-Dist: dataclass-wizard
-Dynamic: author
-Dynamic: author-email
-Dynamic: classifier
-Dynamic: description
-Dynamic: description-content-type
-Dynamic: license-file
-Dynamic: requires-dist
-Dynamic: requires-python
-Dynamic: summary
-
-Helper package for coding and automation
-
-**Version 0.1.67**
-
-+ now use `uv` for venv management
-+ `research/perfcalc`: support both torchmetrics and custom metrics for performance calculation
-
-**Version 0.1.61**
-
-+ add `util/video`: add `VideoUtils` class to handle common video-related tasks
-+ add `util/gpu_mon`: add `GPUMonitor` class to monitor GPU usage and performance
-
-**Version 0.1.59**
-
-+ add `util/perfcalc`: abstract class for performance calculation. This class need to be inherited and implemented with specific performance calculation logic.
-
-**Version 0.1.55**
-
-+ add `util/dataclass_util` to help dynamically create `dataclass` classes from dictionary or YAML file, including support for nested dataclasses. From there, we can use `dataclass_wizard` to create a list of `dataclass` classes with the help from ChatGPT.
-
-**Version 0.1.52**
-
-+ add `research/perftb` module to allow creating and managing performance tables for experiments, including filtering by datasets, metrics, and experiments.
-
-**Version 0.1.50**
-
-+ add `pprint_local_path` to print local path (file/directory) in clickable link (as file URI)
-
-+ add `research` package to help with research tasks, including `benchquery` for benchmarking queries from dataframe
-+ add `wandb` module to allow easy sync offline data to Weights & Biases (wandb) and batch clear wandb runs.
-
-**Version 0.1.47**
-+ add `pprint_box` to print object/string in a box frame (like in `inspect`)
-
-**Version 0.1.46**
-+ filter the warning message of `UserWarning: Unable to import Axes3D.`
-+ auto_wrap_text for `fn_display_df` to avoid long text in the table
-
-**Version 0.1.42**
-+ add <rich_color.py>: add basic color list (for easy usage)
-
-**Version 0.1.41**
-+ add <rich_color.py> to display rich color information in <rich> python package (rcolor_str, rcolor_pallet_all, etc.)
-
-**Version 0.1.40**
-
-+  update <csvfile.py> to use `itables` and `pygwalker` to display dataframe in jupyter notebook.
-
-**Version 0.1.38**
-
-+  add <torchloader.py> to search for best cfg for torch dataloader (num_workers, batch_size, pin_memory, et.)
-
-**Version 0.1.37**
-
-+  add <dataset.py> to help split classification dataset into train/val(test)
----
-**Version 0.1.33**
-
-+ add `plot.py` module to plot DL model training history (with columlns: epoch, train_accuracy, val_accuracy, train_loss, val_loss) using `seaborn` and `matplotlib`
----
-**Version 0.1.29**
-
-+ for `tele_noti` module, `kaleido==0.1.*` is required for plotly since `kaleido 0.2.*` is not working (taking for ever to generate image)
----
-**Version 0.1.24**
-
-+ rename `sys` to `system` to avoid conflict with built-in `sys` module
-+ add `tele_noti` module to send notification to telegram after a specific interval for training progress monitoring
----
-**Version 0.1.22**
-
-+ add `cuda.py` module to check CUDA availability (for both pytorch and tensorflow)
----
-**Version 0.1.21**
-
-+ using `networkx` and `omegaconf` to allow yaml file inheritance and override
----
-**Version 0.1.15**
-
-+ `__init__.py`: add common logging library; also `console_log` decorator to log function (start and end)
-
----
-
-**Version 0.1.10**
-
-+ filesys: fix typo on "is_exit" to "is_exist"
-+ gdrive: now support uploading file to folder and return direct link (shareable link)
-
-**Version 0.1.9**
-
-+ add dependencies requirement.txt
-
-**Version 0.1.8**
-
-Fix bugs:
-
-+ [performance] instead of inserting directly new rows into table dataframe, first insert it into in-memory `row_pool_dict`, that fill data in that dict into the actual dataframe when needed.
-
----
-
-**Version 0.1.7**
-
-Fix bugs:
-
-+ fix insert into table so slow by allowing insert multiple rows at once
-
----
-
-**Version 0.1.6**
-
-New features:
-
-+ add DFCreator for manipulating table (DataFrame) - create, insert row, display, write to file
-
----
-
-**Version 0.1.5**
-
-New Features
-
-+ add cmd module
-+ new package structure
-
----
-
-**Version 0.1.4**
-
-New Features
-
-+ add support to create Bitbucket Project from template
-
----
-
-**Version 0.1.2**
-
-New Features
-
-+ add support to upload local to google drive.
+Metadata-Version: 2.1
+Name: halib
+Version: 0.1.73
+Summary: Small library for common tasks
+Author: Hoang Van Ha
+Author-email: hoangvanhauit@gmail.com
+License: UNKNOWN
+Platform: UNKNOWN
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+License-File: LICENSE.txt
+Requires-Dist: arrow
+Requires-Dist: click
+Requires-Dist: enlighten
+Requires-Dist: kaleido (==0.1.*)
+Requires-Dist: loguru
+Requires-Dist: more-itertools
+Requires-Dist: moviepy
+Requires-Dist: networkx
+Requires-Dist: numpy
+Requires-Dist: omegaconf
+Requires-Dist: opencv-python
+Requires-Dist: pandas
+Requires-Dist: Pillow
+Requires-Dist: Pyarrow
+Requires-Dist: pycurl
+Requires-Dist: python-telegram-bot
+Requires-Dist: requests
+Requires-Dist: rich
+Requires-Dist: scikit-learn
+Requires-Dist: matplotlib
+Requires-Dist: seaborn
+Requires-Dist: plotly
+Requires-Dist: pygwalker
+Requires-Dist: tabulate
+Requires-Dist: itables
+Requires-Dist: timebudget
+Requires-Dist: tqdm
+Requires-Dist: tube-dl
+Requires-Dist: wandb
+Requires-Dist: dataclass-wizard
+
+Helper package for coding and automation
+
+**Version 0.1.73**
+
++ `research/base_exp`: add base experiment class to handle common experiment tasks, including performance calculation and saving results.
+
+**Version 0.1.67**
+
++ now use `uv` for venv management
++ `research/perfcalc`: support both torchmetrics and custom metrics for performance calculation
+
+**Version 0.1.61**
+
++ add `util/video`: add `VideoUtils` class to handle common video-related tasks
++ add `util/gpu_mon`: add `GPUMonitor` class to monitor GPU usage and performance
+
+**Version 0.1.59**
+
++ add `util/perfcalc`: abstract class for performance calculation. This class need to be inherited and implemented with specific performance calculation logic.
+
+**Version 0.1.55**
+
++ add `util/dataclass_util` to help dynamically create `dataclass` classes from dictionary or YAML file, including support for nested dataclasses. From there, we can use `dataclass_wizard` to create a list of `dataclass` classes with the help from ChatGPT.
+
+**Version 0.1.52**
+
++ add `research/perftb` module to allow creating and managing performance tables for experiments, including filtering by datasets, metrics, and experiments.
+
+**Version 0.1.50**
+
++ add `pprint_local_path` to print local path (file/directory) in clickable link (as file URI)
+
++ add `research` package to help with research tasks, including `benchquery` for benchmarking queries from dataframe
++ add `wandb` module to allow easy sync offline data to Weights & Biases (wandb) and batch clear wandb runs.
+
+**Version 0.1.47**
++ add `pprint_box` to print object/string in a box frame (like in `inspect`)
+
+**Version 0.1.46**
++ filter the warning message of `UserWarning: Unable to import Axes3D.`
++ auto_wrap_text for `fn_display_df` to avoid long text in the table
+
+**Version 0.1.42**
++ add <rich_color.py>: add basic color list (for easy usage)
+
+**Version 0.1.41**
++ add <rich_color.py> to display rich color information in <rich> python package (rcolor_str, rcolor_pallet_all, etc.)
+
+**Version 0.1.40**
+
++  update <csvfile.py> to use `itables` and `pygwalker` to display dataframe in jupyter notebook.
+
+**Version 0.1.38**
+
++  add <torchloader.py> to search for best cfg for torch dataloader (num_workers, batch_size, pin_memory, et.)
+
+**Version 0.1.37**
+
++  add <dataset.py> to help split classification dataset into train/val(test)
+---
+**Version 0.1.33**
+
++ add `plot.py` module to plot DL model training history (with columlns: epoch, train_accuracy, val_accuracy, train_loss, val_loss) using `seaborn` and `matplotlib`
+---
+**Version 0.1.29**
+
++ for `tele_noti` module, `kaleido==0.1.*` is required for plotly since `kaleido 0.2.*` is not working (taking for ever to generate image)
+---
+**Version 0.1.24**
+
++ rename `sys` to `system` to avoid conflict with built-in `sys` module
++ add `tele_noti` module to send notification to telegram after a specific interval for training progress monitoring
+---
+**Version 0.1.22**
+
++ add `cuda.py` module to check CUDA availability (for both pytorch and tensorflow)
+---
+**Version 0.1.21**
+
++ using `networkx` and `omegaconf` to allow yaml file inheritance and override
+---
+**Version 0.1.15**
+
++ `__init__.py`: add common logging library; also `console_log` decorator to log function (start and end)
+
+---
+
+**Version 0.1.10**
+
++ filesys: fix typo on "is_exit" to "is_exist"
++ gdrive: now support uploading file to folder and return direct link (shareable link)
+
+**Version 0.1.9**
+
++ add dependencies requirement.txt
+
+**Version 0.1.8**
+
+Fix bugs:
+
++ [performance] instead of inserting directly new rows into table dataframe, first insert it into in-memory `row_pool_dict`, that fill data in that dict into the actual dataframe when needed.
+
+---
+
+**Version 0.1.7**
+
+Fix bugs:
+
++ fix insert into table so slow by allowing insert multiple rows at once
+
+---
+
+**Version 0.1.6**
+
+New features:
+
++ add DFCreator for manipulating table (DataFrame) - create, insert row, display, write to file
+
+---
+
+**Version 0.1.5**
+
+New Features
+
++ add cmd module
++ new package structure
+
+---
+
+**Version 0.1.4**
+
+New Features
+
++ add support to create Bitbucket Project from template
+
+---
+
+**Version 0.1.2**
+
+New Features
+
++ add support to upload local to google drive.
+
+

{halib-0.1.67.dist-info → halib-0.1.73.dist-info}/RECORD

@@ -28,6 +28,8 @@ halib/online/gdrive_mkdir.py,sha256=wSJkQMJCDuS1gxQ2lHQHq_IrJ4xR_SEoPSo9n_2WNFU,
 halib/online/gdrive_test.py,sha256=hMWzz4RqZwETHp4GG4WwVNFfYvFQhp2Boz5t-DqwMo0,1342
 halib/online/projectmake.py,sha256=Zrs96WgXvO4nIrwxnCOletL4aTBge-EoF0r7hpKO1w8,4034
 halib/research/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+halib/research/base_config.py,sha256=AIjVzl2ZJ9b8yIGb2X5EZwLmyGJ_9wNWqrib1nU3Wj0,2831
+halib/research/base_exp.py,sha256=tz03FF2XMI9b6Ram4ZJBjKo053Vb7T8yTTCH0jeLvp4,3218
 halib/research/benchquery.py,sha256=FuKnbWQtCEoRRtJAfN-zaN-jPiO_EzsakmTOMiqi7GQ,4626
 halib/research/dataset.py,sha256=QU0Hr5QFb8_XlvnOMgC9QJGIpwXAZ9lDd0RdQi_QRec,6743
 halib/research/metrics.py,sha256=Xgv0GUGo-o-RJaBOmkRCRpQJaYijF_1xeKkyYU_Bv4U,5249
@@ -49,8 +51,8 @@ halib/utils/gpu_mon.py,sha256=vD41_ZnmPLKguuq9X44SB_vwd9JrblO4BDzHLXZhhFY,2233
 halib/utils/listop.py,sha256=Vpa8_2fI0wySpB2-8sfTBkyi_A4FhoFVVvFiuvW8N64,339
 halib/utils/tele_noti.py,sha256=-4WXZelCA4W9BroapkRyIdUu9cUVrcJJhegnMs_WpGU,5928
 halib/utils/video.py,sha256=ZqzNVPgc1RZr_T0OlHvZ6SzyBpL7O27LtB86JMbBuR0,3059
-halib-0.1.67.dist-info/licenses/LICENSE.txt,sha256=qZssdna4aETiR8znYsShUjidu-U4jUT9Q-EWNlZ9yBQ,1100
-halib-0.1.67.dist-info/METADATA,sha256=Zk22ct5W95qBzGkz0tNepuAdfUwPJTbVO7Nb4L_hFTQ,5541
-halib-0.1.67.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-halib-0.1.67.dist-info/top_level.txt,sha256=7AD6PLaQTreE0Fn44mdZsoHBe_Zdd7GUmjsWPyQ7I-k,6
-halib-0.1.67.dist-info/RECORD,,
+halib-0.1.73.dist-info/LICENSE.txt,sha256=qZssdna4aETiR8znYsShUjidu-U4jUT9Q-EWNlZ9yBQ,1100
+halib-0.1.73.dist-info/METADATA,sha256=rbQYLf3CgTFzNtQ66v6OCEHpE9yjNfT2Qf7LDd-II-g,5350
+halib-0.1.73.dist-info/WHEEL,sha256=G16H4A3IeoQmnOrYV4ueZGKSjhipXx8zc8nu9FGlvMA,92
+halib-0.1.73.dist-info/top_level.txt,sha256=7AD6PLaQTreE0Fn44mdZsoHBe_Zdd7GUmjsWPyQ7I-k,6
+halib-0.1.73.dist-info/RECORD,,

{halib-0.1.67.dist-info → halib-0.1.73.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.9.0)
+Generator: bdist_wheel (0.37.1)
 Root-Is-Purelib: true
 Tag: py3-none-any