HotFix: v0.8.1 - Pancake

NadimGhaznavi released this 13 Oct 20:03

· 53 commits to main since this release

a182115

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog
and this project adheres to Semantic Versioning.

[0.8.1] - 2025-10-13

Added

Show the number of frames for the long training phase, when mem_type is random frames.

Fixed

Increased width so Random Frames isn't truncated.
Fixed SQL bug in ReplayMemory:get_random_frames().

Changed

Changed the highscores layout so there's a space between the scrollbar and the seconds.
Set the stored games back to zero when the restart button is pressed
Clear the data from the games and frames SQLite3 tables when the restart button is pressed.
Format the number of stored games (add commas) when they are large.

[0.8.0] - 2025-10-13

Changed

Linear Model, ModelL, changes:
- Formatting improvement of the highscores.
- Increased the linear model's (ModelL) learning rate.
- Added an additional hidden layer to the linear model.
- Decreased the dropout value from 0.2 to 0.1.

[0.7.0] - 2025-10-13

Added

Added a Time column to the Highscores
A screenshot of the TUI to the website.

Changed

Plotting Widget:
- Renamed the plotting widget from Db4EPlot to LabPlot
- Removed the code that averaged results since I'm using a sliding method instead.
Added a target model that keeps a frozen copy of the main network. This prevents chasing moving targets instability issues.
Added a soft update (τ=0.01) to blend weights in slowly. This smooths the learning curve.
Replaced MSE with Huber loss. It's less sensitive to large TD errors (outliers).
Added gradient clipping to prevents exploding gradients.
Added an update frequency (target_update_freq=100) to sync target every ~100 frames
Vectorize the data coming out of the ReplayMemory leading to a HUGE improvement in speed.
Adjust gradients on the batch, not single frames for better learning and smoother gradients
Return the loss for future plotting.

Fixed

Cleared the Plot Widget when the Restart button is pressed
Reset the neural network model's learned weights when the Restart button is pressed.

[0.6.0] - 2025-10-12

Added

Added a DColors constants file.
Updated website content.
- Added Schema, Constants, Project Layout, Git Branching and Git Commit technical documents.
- Added a Richard Sutton page.
ReplayMemory exposes the current training game ID:
- Modified the ReplayMemory to return the stored game and frame IDs.
- Modified the AIAgent to first load the training data, get the training game ID, then train.
- Modified AISim to get the training game ID from teh AIAgent and display it in the TUI.

Changed

Set the learning rate in the AITrainer, based on the selected model.

Refactor

Made handling and definition of constants consistent.
Renamed the DDb4ePlot constants file to DPlot

[0.5.0] - 2025-10-11

Added

Added a highscores widget showing the game number and the high score.
Reworked the state map in GameBoard:get_state()
Moved the ModelRNN model parameters into the DModelRNN constants file.
Updated the SQLite schema used by the ReplayMemory class.
- See https://snakelab.osoyalce.com/db_schema.html
Added drop down menu to select the ReplayMemory type i.e. Random Game, Random Frames or None.
- Added a runtime widget to display the current memory type
Added sections to the website / README file.
Added a Richard Sutton page.

[0.4.8] - 2025-10-11

Fixed

Fix path to AISim.tcss.

[0.4.7] - 2025-10-11

Changed

Use tempfile module to get a temporary file location for SQLite3 database.
Use safer cleanup for SQLite3 database.

[0.4.6] - 2025-10-11

Fixed

Fix path issue with ReplayMemory's use of SQLite.

[0.4.6] - 2025-10-11

Fixed

Add main() function to AISim.py so the CLI entry point ai-snake-lab works after a pip install ai-snake-lab.

[0.4.5] - 2025-10-11

Fixed

Fix import statements in ai/* and game/* files.

[0.4.4] - 2025-10-11

Fixed

Fix import statements in Constants files.

[0.4.3] - 2025-10-11

Added

Initial PyPI release.

Fixed

Changed

Assets 2