Closed Oufattole closed 1 month ago
The recent changes enhance the configuration and code modules across the project, focusing on optimizing testing efficiency, refining model performance metrics, and improving training methodologies. Notable updates include the exclusion of slow tests from pytest runs, a shift in monitoring metrics to prioritize validation loss, and new configurations for Ray Tune and transfer learning. These modifications aim to streamline workflows, enhance evaluation accuracy, and improve the overall training process.
Files | Change Summary |
---|---|
.github/workflows/tests.yaml |
Updated pytest command to exclude slow tests for faster feedback. |
configs/callbacks/*.yaml |
Shifted monitoring metrics from validation accuracy to validation loss. |
configs/debug/limit.yaml |
Increased limit_train_batches from 0.01 to 0.05 . |
configs/eval.yaml |
Renamed task_name to name , now explicitly set to null . |
configs/experiment/example.yaml |
Changed dataset from "mnist" to "pytorch_dataset" and model type from "mnist" to "supervised." |
configs/hparams_search/*.yaml |
Updated optimization metrics; added/remodeled hyperparameters. |
configs/model/token_forecasting.yaml |
Changed initialization target from TripletTokenForecastingModule to TokenForecastingModule . |
configs/paths/default.yaml |
Changed root_dir from environment variable to "." . |
configs/train.yaml |
Added new name field set to "train" . |
configs/trainer/ray.yaml |
Defined configurations for training environment using Ray and Lightning. |
configs/transfer_learning.yaml |
Created structured framework for transfer learning settings. |
pyproject.toml |
Added ray[tune] dependency and new script entries for transfer and tuning. |
Various src/meds_torch/*.py |
Introduced new functionalities, removed redundancy, and improved methods across multiple files. |
tests/* |
Restructured test files significantly, introducing new tests and modifying existing ones for clarity. |
sequenceDiagram
participant User
participant TestRunner
participant Model
participant Config
User->>TestRunner: Execute Tests
TestRunner->>Config: Load Configuration
Config->>Model: Set Parameters
Model-->>Config: Parameters set
TestRunner->>Model: Run Tests
Model-->>TestRunner: Return Results
TestRunner->>User: Display Results
π In fields so wide and skies so blue,
I twirl in joy for changes new!
With tests that run and metrics bright,
Our models learn to reach new height.
So hop along, let's celebrate,
These tweaks will surely make us great! π
Thank you for using CodeRabbit. We offer it for free to the OSS community and would appreciate your support in helping us grow. If you find it useful, would you consider giving us a shout-out on your favorite social media?
Summary by CodeRabbit
New Features
Improvements
Bug Fixes