salesforce_ai-dl-ml-llm_github

Salesforce Einstein

Introduced in 2016, Salesforce Einstein is an integrated set of AI technologies embedded within the Salesforce platform. It leverages machine learning, deep learning, predictive analytics, and natural language processing to deliver smart data discovery and actionable insights. Developers can access various APIs to build custom AI-powered applications, enhancing customer relationship management through intelligent automation and analytics.

https://github.com/SalesforceSFDC/Einstein

Salesforce XGen

In 2023, Salesforce AI Research released XGen, a family of Large Language Models (LLMs) with 7 billion parameters, trained on sequences up to 8,000 tokens. The XGen models are designed for long-sequence modeling, enabling advanced natural language processing tasks such as document summarization and code generation. This development enhances the capability to handle extensive textual data efficiently.

https://github.com/salesforce/xgen

CodeRL

CodeRL, introduced in 2022, is a framework that combines reinforcement learning with code generation techniques. It utilizes the CodeT5 family of encoder-decoder language models, pre-trained on extensive datasets with improved learning objectives. The framework aims to enhance the quality and efficiency of automated code synthesis, contributing to advancements in AI-assisted programming.

https://github.com/salesforce/CodeRL

OmniXAI

Launched in 2022, OmniXAI (Omni eXplainable AI) is a Python library offering comprehensive explainable AI and interpretable machine learning capabilities. It supports various data types—including tabular data, images, texts, and time-series—and integrates with multiple ML models from Scikit-learn, PyTorch, and TensorFlow. OmniXAI provides diverse explanation methods, facilitating a deeper understanding of model decisions across different stages of the ML process.

https://github.com/salesforce/OmniXAI

ml4ir

Introduced in 2021, ml4ir (Machine Learning for Information Retrieval) is a framework designed to apply machine learning techniques to information retrieval tasks. It enables the development of models that improve search relevance and ranking, leveraging deep learning architectures to enhance the retrieval process. ml4ir supports scalable deployment, making it suitable for large-scale search applications.

https://github.com/salesforce/ml4ir

WarpDrive

Released in 2021, WarpDrive is an open-source framework that facilitates end-to-end deep multi-agent reinforcement learning on a single or multiple GPUs. By leveraging the parallelization capabilities of GPUs, WarpDrive achieves significant speedups in training RL models, enabling efficient simulation and learning across numerous agents and environment replicas.

https://github.com/salesforce/warp-drive

DeepTIMe

In 2023, Salesforce introduced DeepTIMe, a PyTorch-based model for time-series forecasting. It employs a meta-optimization formulation to learn deep time-index representations, achieving competitive results with state-of-the-art methods in long-sequence time-series forecasting. DeepTIMe is designed for efficiency, making it suitable for real-world applications requiring accurate temporal predictions.

https://github.com/salesforce/DeepTIMe

decaNLP

Launched in 2018, decaNLP (The Natural Language Decathlon) is a framework that unifies ten natural language processing tasks into a single multitask learning setup. It introduces the Multitask Question Answering Network (MQAN), a model that jointly learns all tasks without task-specific modules or parameters. This approach facilitates transfer learning and zero-shot evaluation across diverse NLP tasks.

https://github.com/salesforce/decaNLP

CodeGen2

In 2023, Salesforce released CodeGen2, a series of Large Language Models trained on both programming and natural languages. These models are designed to assist in code generation and understanding, providing developers with tools to automate coding tasks and enhance software development efficiency. The CodeGen2 models are available in various parameter sizes, catering to different application needs.

https://github.com/salesforce/CodeGen

LangChain

Introduced in 2022, LangChain is a software framework that facilitates the integration of Large Language Models into applications. It supports use cases such as document analysis, summarization, chatbots, and code analysis. LangChain provides modular tools for building advanced AI-driven systems, enabling developers to create complex LLM-based applications efficiently.

https://github.com/langchain-ai/langchain

salesforce_ai-dl-ml-llm_github.txt · Last modified: 2025/02/01 06:30 by 127.0.0.1

Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki