uptrain

Kagura AI is an AI-powered testing tool designed for modern development teams. It generates, executes, and evolves QA tests automatically from a URL and a description of what to test. Kagura AI adapts to UI changes, allowing users to pause mid-test to interact, eliminating scripting and maintenance.

TransformerLens is a library for mechanistic interpretability of GPT-style language models. It allows researchers and engineers to analyze the internal workings of transformer models. The tool supports in-depth investigation of model behavior.

transformers-interpret is a model explainability tool designed for use with the 🤗 transformers package. It allows users to explain their transformer models in just two lines of code. The tool provides explainers for various model types, enhancing transparency and understanding of model predictions. It aids in debugging and improving model performance.

Lm-evaluation-harness is a framework designed for few-shot evaluation of language models. It allows researchers and engineers to assess model performance across various tasks. The tool supports CLI refactoring with subcommands and YAML config files. It also offers lighter installation options with separate model backends.

Coderabbit is an AI-powered pull request reviewer. It provides context-aware feedback and line-by-line code suggestions. The tool offers real-time chat for collaboration. Coderabbit aims to cut code review time and bugs in half for fast-moving teams.

ttach is a PyTorch library for image test time augmentation. It performs random modifications to test images to improve model performance. The tool averages the predictions of each corresponding augmented image. ttach is open-source and available on GitHub.

Shapash is a Python library that enhances machine learning interpretability. It offers visualizations with clear labels, making it easier to understand model features and interactions. Shapash simplifies the comprehension of machine learning models for users of all levels. It generates a web application for seamless navigation between local features.

tide is a toolbox for identifying object detection errors. It helps compute and evaluate the effect of object detection and instance segmentation on overall performance. The tool is designed to be easy to use and provides insights into improving object detection models.

TextAttack is a Python framework for adversarial attacks, data augmentation, and model training in NLP. It allows users to generate adversarial examples to test and improve the robustness of NLP models. The tool is open-source and provides documentation on ReadTheDocs.