Product library

Open source chat robot with performance close to ChatGPT

Vicuna is an open source chat robot, which conducts fine tuning training for LLaMA through user shared conversations. The preliminary assessment uses GPT-4 as the judge, which shows that Vicuna-13B has reached 90% * quality of OpenAI ChatGPT and Google Bard when it exceeds 90% *, and outperforms other models such as LLaMA and Stanford Alpaca when it exceeds 90% *. The training cost of Vicuna-13B is about $300. Code and model weights and online presentations are available for non-commercial use.

The local chat robot is finely trained by Bilibili user comments

Bilibot is a local chat robot based on Bilibili user comment training, which supports text chat and voice conversation. It uses Qwen1.5-32B-Chat as the basic model and fine tunes it in combination with Apple's mlx lm LORA project. The voice generation part is based on the GPT-SoVITS project and uses the Pemont voice model. The robot can quickly generate dialogue content, which is suitable for occasions where an intelligent dialogue system is required.

Open source framework for building voice and multimodal dialogue AI.

Pipecat is an open source framework for building voice and multimodal dialogue agents, such as personal trainers, meeting assistants, children's story toys, customer support robots, reception processes and smart social partners. It supports local operation, can be migrated to the cloud, integrates a variety of AI services and transmission modes, and is highly customizable and extensible.

Prompto is an open source network application designed to make interaction with LLM simple and efficient.

Prompto is an open source web application designed to make interaction with large language models (LLMs) simple and efficient. It can easily switch between different LLMs, adjust the creativity and risk level of LLM by adjusting the temperature setting, provide chat robot interface and notebook interface, create templates for common prompts, and run in the browser to ensure a smooth response experience. All settings and chat records are only stored in the browser's local storage to protect user privacy.

Provide the most advanced machine learning models, data sets and tools for real world robots.

LeRobot is an open source project aimed at lowering the threshold of entering the robot field, allowing everyone to contribute and benefit from shared data sets and pre training models. It contains the most advanced methods that have been verified in the real world, with special attention to imitation learning and reinforcement learning. LeRobot provides a set of pre training models, data sets with human collection demonstration and simulation environment so that users can start without assembling robots. In the coming weeks, it is planned to increase support for the most affordable and capable real world robots.

A corpus composed of real world users interacting with ChatGPT.

The WildChat dataset is a corpus composed of 1 million real world users interacting with ChatGPT, characterized by diversity of languages and user prompts. This data set is used to fine tune the Llama-2 of the Meta, and created the WildLlama-7b-user-asistant chat robot, which can predict user prompts and assistant responses.

An open source chat robot that can explain concepts, write poems, program, and solve logical puzzles.

Chat With Llama 3 is an open source chat robot developed by Meta AI. It can carry out a variety of intelligent dialogues, including explaining complex concepts, creating poetry, writing code, solving logical puzzles, and even helping users name pets. The main advantage of this chat robot is its versatility and open source, which can be widely used in various scenarios, and can be customized and improved according to needs.

Lightweight ChatGPT alternatives designed for team collaboration

Joia is a lightweight ChatGPT alternative designed for team collaboration, focusing on privacy protection and open source. It allows users to easily provide AI chat access to everyone in the organization, create a special-purpose chat robot and share it with the team. Joia directly links with large language model providers such as OpenAI through API, and only pays according to the token actually used. Compared with ChatGPT's pricing plan, it can save up to 75% of the subscription cost.

Create and use a custom chat robot based on the open source model of HuggingFace.

HuggingChat Assistants is a customized platform for chat robots released by HuggingFace. Users can select multiple open source models hosted by HuggingFace to create customized chat robots, which are applicable to multiple fields.

Miqu 1-70b is an open source large-scale language model

Miqu 1-70b is an open source large-scale language model, which adopts a novel self-monitoring learning method and can handle various natural language tasks. The model has 17 billion parameters, supports multiple prompt formats, and can fine tuning to generate high-quality text. Its powerful ability of understanding and generating makes it widely used in chat robots, text summaries, question answering systems and other fields.

AIWaves is an AI software company that provides AI products and services.

AIWaves is an AI startup headquartered in Hangzhou, which focuses on applying large-scale language models to Internet scenarios. With the long-term mission of "realizing AGI and making human life better" and the vision of becoming "the world's first AI content platform", the company is committed to redefining the future content creation and consumption paradigm. The company's main products include: dialogue robot waveform Bot, AI writing assistant frog story, open source toolkit Agents, etc. These products can be widely used in e-commerce, games, medical care, education and other fields to provide users with better content and services.

Generated AI service of Oracle cloud platform

Oracle Generic AI Service is a fully hosted generative AI service, which can easily integrate multi-functional language models into various use cases, including writing assistance, summary generation and chat robots. Based on Oracle Cloud Infrastructure, this service provides a high-performance, enterprise oriented model, supports efficient fine-tuning and multiple use cases, and provides secure and reliable enterprise level data management, security and control.

Multi language pre training language model

InternLM2, "Scholar Puyu 2.0", is a large multilingual pre training language model for Chinese and English. It has powerful capabilities such as language understanding, natural language generation, multi-mode reasoning, code understanding, etc. The model adopts Transformer architecture and carries out massive data pre training, which has reached the industry leading level in many directions, such as long text understanding, dialogue, mathematical operation, etc. This series of models includes a variety of scales, and users can select appropriate models to fine tune downstream tasks or build applications such as chat robots.

Open source Chinese English bilingual pre training language model

LingoWhale-8B is an open source large-scale bilingual pre training language model with strong natural language understanding and generation capabilities. Through pre training on massive high-quality Chinese and English data, it can complete the understanding of long texts and multiple rounds of interaction. The model adopts Transformer architecture, with a parameter amount of 8 billion. It has achieved leading results in many Chinese and English public benchmark tests. LingoWhale-8B is fully open to academic research, and individual developers can use it for commercial purposes for free. This model can be widely used in chat robot, knowledge question and answer, text generation and other fields.

Build an open source platform for chat robots

OpenAssistantGPT is an open source platform for building chat robots using OpenAI assistants. It uses OpenAI assistants to power our chat robots. You can use GPT 4 or GPT 3.5. Just include our js code in your website, which is cheap. OpenAssistant GPT is completely open source, and you can find its code on GitHub. You can learn how to implement our chat robot in your website by viewing our documents.

Build your chat robot

Botticelli is an open source NET Core framework, which is used to build your own general chat robot and integrates database, queue agent, voice engine and AI engine, such as GPT-3 and ChatGPT. Botticelli is cross platform, reliable, easy to use, easy to deploy, supports Docker, easy to integrate with databases, AI solutions, voice synthesizers and queue agents, and can be used for scheduling high load systems.

Open source chat robot framework, supporting rapid deployment of private ChatGPT applications

LobeChat is an open source scalable high-performance chat robot framework, which supports the free deployment of private ChatGPT/LLM network applications with one click. With functions such as user-defined model, multi language support, Plugins system, knowledge extraction, etc., it can help users quickly build private, safe and controllable AI assistants and knowledge management tools.

Build AI chat robot and AI driven text area in the application

CopilotKit is an open source Copilot platform that can build AI chat robots and AI driven text areas in React Web applications. It combines the front-end SDK, back-end SDK and optional cloud infrastructure to provide a comprehensive solution.

Intelligent document assistant, making document processing more efficient

DocsAI is an intelligent document assistant, which can train documents, communicate with documents, and create chat robots to solve user problems. Users can easily find any content in the document, add documents from multiple sources, customize the appearance of the assistant, put forward answers and suggestions, use custom APIs for integration, and embed chat windows on the website. In addition, it also supports access to dialogue summaries, user sentiment and clue identification, mobile device friendliness, export of dialogue records, multilingual processing, and open source.

The first choice of modern customer service team

Chatroot is an open source AI customer service platform, which improves customer support experience through AI technology, provides self-service tools, global search interface, real-time chat SDK and other functions, and supports self hosting deployment. It conforms to SOC 2 Type II standard, and can seamlessly switch between different large-scale language models. As an alternative to Intercom, Zendesk and Salesforce Service Cloud, Chatboot provides functions such as shared inbox, multi-channel support, automation, chat robot, etc.

Generating robot simulation tasks using large language models

GenSim uses large language models to generate a large number of robot simulation tasks, and supports two modes of goal oriented generation and exploratory generation, which can be used for multi task strategy training and task level generalization. The use of GPT4 has expanded the existing benchmark by more than 10 times, supporting more than 100 tasks. Through supervised fine-tuning and evaluation of multiple LLMs, including fine-tuning GPT and Code Llama, code for robot simulation tasks is generated. After the smallest simulation is adapted to the real world, the multi task strategy pre trained on the simulation task generated by GPT4 shows a stronger transfer ability in the real world, 25% more than the baseline.

Open source GPT architecture AI chat robot

SynthIA-7B-v1.3 is an open source chat robot model based on GPT-3 architecture. It can carry out natural language dialogue for a long time, and has a strong ability to understand and generate. It can be used for various applications requiring language interaction to provide a real and intelligent interactive experience.

WordPress AI plug-in, adding chat robot, AI form, AI partner, content generation and other functions

AI Engine is a plug-in that provides AI functions for WordPress. It can add chat robots, AI forms, AI partners, content generation and other functions. Through AI Engine, you can easily generate content, try various tasks (such as translation and correction), add ChatGPT style chat robots to your website, track AI usage, set restrictions, etc. in our interesting AI Playground! You can also create datasets and fine tune them to make your site better. And don't worry, we provide APIs for other plug-ins. Come and try it and see the magic effect. More importantly, enjoy it:)

Easier learning Islam, AI power learning robot, quoted from the Koran and Hadith

Salam Chat is a chat robot based on artificial intelligence, which aims to learn Islam more easily. It provides quotation and citation of Islamic knowledge, and has halal filtering function. Users can get references to the Koran and Hadith by interacting with robots to further understand Islam. This product is trained by many Islamic scholars to ensure that the correct information is recommended. Salam Chat is a free and open source project developed and maintained by the Minority Think Tank Foundation.

The first open source chat GPT

OpenChatKit is an open source project, which provides a strong foundation for various applications to create dedicated and universal chat robots. It includes a large language model to guide tuning, a user-defined tuning template, an extensible retrieval system, and a content filtering model. OpenChatKit can handle various natural language tasks such as dialogue, question and answer, classification, extraction, summarization, etc. It is based on the OIG-43M dataset created by Together, LAION and Ontocord for training.

Browser extension using multiple chat robots at the same time

ChatHub is a browser plug-in that allows you to use multiple chat robots at the same time. It supports 10+open source chat robots such as ChatGPT, Bing, Google Bard, Claude, etc. ChatHub provides most of the functions used for free, as well as the paid ChatHub Premium package, which unlocks all advanced functions.

ChatGPT AI chat robot based on open source and free software

AINIRO provides ChatGPT AI chat robot, AI search and AI expert system based on open source and free software. Our ChatGPT chat robot can create customized machine learning models based on your website content for customer service, e-commerce, potential customer generation, etc. Our robots can also integrate with WhatsApp, Messenger and SMS, and even with search engines or local SQL databases.

The first platform for creating robots driven by AI and Web3

MyShell is an innovative platform for creating personalized chat robots. Here, you can create a practical and entertaining robot called Shell through an interactive workshop, which is not only suitable for personal use, but also can be shared with friends and communities. The platform provides customizable functions such as knowledge integration, voice studio, unlimited memory, Internet and API integration, multimodal integration, and customized data fine-tuning. Please check the official website for price and positioning.