skleaarn 1.0.0

ML competition problem solutions and study notes collection

skleaarn

A personal ML problems collection and learning notes package. Installing this package automatically copies all ML competition solutions and study notes to your Desktop.

Installation

pip install skleaarn

On first import, the package will automatically install the Sklearn folder to your Desktop (or ~/Documents/Sklearn as fallback). It also creates a Start Menu shortcut on Windows.

Manual Install

import skleaarn
skleaarn.install()                     # installs to ~/Desktop/Sklearn
skleaarn.install("C:/MyFolder")       # installs to a custom location

Or via command line:

skleaarn                               # installs to ~/Desktop/Sklearn
skleaarn C:/MyFolder                  # installs to a custom location

What's Included

Problems/

58 Python files with ML competition solutions, covering:

• Computer Vision: UNet segmentation, ResNet classification, GhostHunting, HotSpot, Skeletons, EmojiSegmentation, Broken image tasks, GlitchHunter
• NLP / Text: ToxicOnline (BiLSTM + FastText + Word2Vec), HieroglyphHunter (GloVe), OmVsAI, SavingChristmas
• Tabular ML: HeartBeat, RaspberryPicking, ANPC, CalitateaSolului, Churn, Muzee, FaultyLLM
• Reinforcement Learning: RL Value Iteration, Q-Learning (LandoNorris, MountainCar), Drone RL, Labirint RL, StochasticRIFT
• Audio: AICC3 (sound classification), DigitTags

LearnIt_Notes/

• NOTES.md — 31 categorized theory notes covering: general PyTorch patterns, ResNet, LSTM/BiLSTM, NLP (GloVe, Word2Index, Padding), UNet, Computer Vision tricks, Audio (spectrograms), important gotchas
• snippets/ — 5 ready-to-use code templates:
  • 01_Scheduler_When_and_What.py
  • 02_F1_Train_Val_loop.py
  • 03_UNET.py
  • 04_Double_Conv_for_UNET.py
  • 05_Train_Val_split.py

Categories Covered

Category	Description
general	PyTorch patterns, loss functions, model saving
resnet	ResNet fine-tuning, embeddings, unfreezing layers
lstm	BiLSTM, multi-label, Word2Index
nlp	GloVe, sentence embeddings, padding, NLTK
vision	UNet, ConnectedComponents, image preprocessing
audio	Spectrograms, audio segmentation
unet	UNet getitem, prediction loops
important	Critical reminders (zero_grad, getting predictions)
output	Cosine similarity for retrieval

Version

1.0.0 — Initial release