3 Repos
Utilities for instantiating neural networks with specific architectures, loss functions, and training configurations.
Distinct from Neural Network Initializers: Existing candidates were either too specific to language models or focused only on weight value initialization.
Explore 3 awesome GitHub repositories matching artificial intelligence & ml · Model Initializers. Refine with filters or upvote what's useful.
This project is a PyTorch person re-identification framework designed for training and evaluating models that identify individuals across different camera views. It provides a complete model training pipeline, a deep learning feature extractor for converting images into numeric vectors, and a suite of computer vision benchmarking tools to measure identity retrieval accuracy. The framework includes a specialized transfer learning toolkit that supports layer freezing, staged learning rate optimization, and differential learning rates for fine-tuning pretrained models. It distinguishes itself th
Creates neural network instances by specifying architecture, training identities, and loss functions.
This project is a collection of educational resources and reference implementations for neural network development using TensorFlow. It serves as a comprehensive learning course, machine learning curriculum, and practical implementation guide for building deep learning architectures. The codebase provides instructional materials and examples covering a wide range of model types, including convolutional neural networks for image classification, recurrent networks and long short-term memory cells for sequential data, and autoencoders for generative modeling. It also includes implementations for
Evaluates how different starting values for variables affect the model's ability to reach a global optimum.
Grok is a neural network training framework and machine learning experiment suite designed for algorithmic generalization research. It provides a set of tools to study how neural networks transition from memorizing training data to discovering general rules when trained on small datasets. The implementation focuses on deep learning overfitting analysis and neural network training evaluation. It enables the execution of training loops to observe the phenomenon of grokking and measure model performance on unseen algorithmic data. The codebase covers capability areas including algorithmic datas
Implements symmetric random weight distributions to ensure consistent convergence across experimental runs.