3 repositorios
Awesome GitHub RepositoriesMulti-Modal Data Processing
Tools and interfaces for managing and preprocessing diverse media types such as text, image, audio, and video for AI training.
Distinct from Cross-Modal Context Management: None of the candidates cover the general CLI-based processing of multiple modalities before training; they focus on retrieval, binding, or context management.
Explore 3 awesome GitHub repositories matching artificial intelligence & ml · Multi-Modal Data Processing. Refine with filters or upvote what's useful.
Awesome Multi-Modal Data Processing GitHub Repositories
Encuentra los mejores repositorios con IA.Buscaremos los repositorios que mejor coincidan usando IA.
- pku-alignment/align-anything
PKU-Alignment/align-anything
4,661Ver en GitHubAlign-anything es un framework de alineación para modelos de lenguaje grandes (LLM) multimodales, diseñado para ajustar modelos en texto, imagen, video y audio. Funciona como un orquestador de entrenamiento distribuido y un kit de herramientas para implementar aprendizaje basado en preferencias, asegurando que el comportamiento del modelo coincida con las intenciones y valores humanos. El framework proporciona pipelines especializados para Supervised Fine-Tuning y Direct Preference Optimization. Incluye un wrapper de motor de inferencia de alto rendimiento para modelos actor, reduciendo el tiempo de generación de secuencias, y un entorno de entrenamiento dedicado para refinar modelos de visión-lenguaje-acción utilizados en robótica física. El sistema gestiona el procesamiento de datos multimodales mediante una interfaz de línea de comandos y soporta el despliegue automatizado de cargas de trabajo de entrenamiento en clusters de hardware con gestión de recursos. Sus capacidades cubren la implementación de algoritmos de alineación, fine-tuning multimodal y optimización de recursos de hardware.
Includes a command-line interface to manage and streamline the processing of diverse media inputs before training.
Python - flagai-open/flagai
FlagAI-Open/FlagAI
3,870Ver en GitHubFlagAI is a distributed deep learning framework and platform designed for the end-to-end lifecycle of large-scale foundation models. It provides a toolkit for training, fine-tuning, and deploying large language models and multi-modal systems across multi-node computing clusters. The project features hardware-agnostic compute abstractions to ensure consistent execution across different accelerators. It includes a dedicated library for parameter-efficient fine-tuning, allowing large neural networks to be adapted to specific tasks with minimal parameter updates and reduced computational overhead
Provides a unified execution interface for processing diverse data types like text and images.
Python - open-mmlab/mmpretrain
open-mmlab/mmpretrain
3,842Ver en GitHubmmpretrain is a modular PyTorch computer vision framework designed for developing, training, and benchmarking deep learning architectures. It serves as a comprehensive toolkit for vision tasks, providing a specialized platform for multimodal machine learning and self-supervised learning. The project features a computer vision model zoo containing architectural definitions and pre-trained weights for backbones such as ViT, ConvNeXt, and Swin Transformer. It distinguishes itself through a dedicated self-supervised learning toolkit that implements algorithms like MAE and DINO to train models wit
Processes diverse media types through specialized encoders and shared embedding spaces for joint image and text analysis.
Pythonbeitclipconstrastive-learning
