Predibase
Declarative machine learning systems provide the best of flexibility and simplicity to enable the fastest-way to operationalize state-of-the-art models. Users focus on specifying the “what”, and the system figures out the “how”. Start with smart defaults, but iterate on parameters as much as you’d like down to the level of code. Our team pioneered declarative machine learning systems in industry, with Ludwig at Uber and Overton at Apple. Choose from our menu of prebuilt data connectors that support your databases, data warehouses, lakehouses, and object storage. Train state-of-the-art deep learning models without the pain of managing infrastructure. Automated Machine Learning that strikes the balance of flexibility and control, all in a declarative fashion. With a declarative approach, finally train and deploy models as quickly as you want.
Learn more
MLBox
MLBox is a powerful Automated Machine Learning python library. It provides the following features fast reading and distributed data preprocessing/cleaning/formatting, highly robust feature selection and leak detection, accurate hyper-parameter optimization in high-dimensional space, state-of-the art predictive models for classification and regression (Deep Learning, Stacking, LightGBM), and prediction with models interpretation. MLBox main package contains 3 sub-packages: preprocessing, optimization and prediction. Each one of them are respectively aimed at reading and preprocessing data, testing or optimizing a wide range of learners and predicting the target on a test dataset.
Learn more
Axolotl
Axolotl is an open source tool designed to streamline the fine-tuning of various AI models, offering support for multiple configurations and architectures. It enables users to train models, supporting methods like full fine-tuning, LoRA, QLoRA, ReLoRA, and GPTQ. Users can customize configurations using simple YAML files or command-line interface overrides, and load different dataset formats, including custom or pre-tokenized datasets. Axolotl integrates with technologies like xFormers, Flash Attention, Liger kernel, RoPE scaling, and multipacking, and works with single or multiple GPUs via Fully Sharded Data Parallel (FSDP) or DeepSpeed. It can be run locally or on the cloud using Docker and supports logging results and checkpoints to several platforms. It is designed to make fine-tuning AI models friendly, fast, and fun, without sacrificing functionality or scale.
Learn more
Automaton AI
With Automaton AI’s ADVIT, create, manage and develop high-quality training data and DNN models all in one place. Optimize the data automatically and prepare it for each phase of the computer vision pipeline. Automate the data labeling processes and streamline data pipelines in-house. Manage the structured and unstructured video/image/text datasets in runtime and perform automatic functions that refine your data in preparation for each step of the deep learning pipeline. Upon accurate data labeling and QA, you can train your own model. DNN training needs hyperparameter tuning like batch size, learning, rate, etc. Optimize and transfer learning on trained models to increase accuracy. Post-training, take the model to production. ADVIT also does model versioning. Model development and accuracy parameters can be tracked in run-time. Increase the model accuracy with a pre-trained DNN model for auto-labeling.
Learn more
