Deep-ML-Solutions

This repository contains fully native Python / NumPy implementations for the deep learning and machine learning problems found on Deep-ML.

We have successfully implemented 118 key problems covering:

• Linear Algebra & Decompositions (SVD, Eigenvalues, Orthogonal Projections)
• Machine Learning & Metrics (Regression, SVM, K-Means, K-Fold, F-Score, Precision/Recall, Clustering)
• Computer Vision (ResNet Blocks, DenseNet Blocks, Normalization techniques, Fractional Pooling, MDN)
• Transformer & Attention Mechanics (Multi-Head / Sliding Window / Grouped Query Attentions, KV Caching, Positional Encodings)
• Advanced LLM Scaling & RL (LoRA, GRPO, FlashAttention, SwiGLU, MoE Gating, Quantization (INT8/FP8/MXFP4), Roofline Profiling)

Problem List

For the complete, detailed list of all solved problems, please see the autogenerated document here: 👉 Deep_ML_Problems_List.md

Testing

All solutions are equipped with local validation logic to ensure correct shape broadcasting, mathematical accuracy, and strict numerical adherence to the Deep-ML spec. Tests are located in the numpy/tests/ directory.

---

Created automatically during an AI-assisted deep-learning problem-solving marathon.