# Auto-GPT 0.4.1 Release Bulletin ## Introduction Welcome to the 0.4.1 Release! In this bulletin, we will be summarizing the most recent commit changes made by various contributors. These updates focus on improving command reliability, fixing bugs, and introducing new features. Let's dive in and explore the details of these changes. ## Features ### Directory-based plugins One of the major highlights of this update is the introduction of the directory-based plugins feature. Previously, plugins had to be compressed into zip files, but now, simply placing a plugin in a subfolder or repository clone within the plugins directory will allow it to be recognized and loaded. This change provides added convenience for developers and simplifies the plugin integration process. ### Keyboard navigation To cater to users who prefer keyboard navigation over mouse interaction, Auto-GPT now supports keyboard navigation in the command line interface (CLI). This improvement enhances the overall user experience and makes navigation more efficient. ### JSON fixes Several issues related to prompts and JSON handling have been addressed in this update. These fixes ensure a more reliable selection and execution of commands, enhancing the stability of Auto-GPT. ### More dependable search Auto-GPT's internet searching functionality has been improved to be more reliable and resilient against errors. The tool now automatically retries searches when it encounters duckduckgo rate limits, ensuring a more dependable search experience. ### Better read_file Building on the enhancements introduced in the previous version (v0.4.0), Auto-GPT now further improves its reading and writing of files. These improvements enhance the accuracy and efficiency of file-related operations. ### More reliable Python code execution Auto-GPT's ability to create and execute Python code has been enhanced in this update, resulting in a more reliable and smoother experience. Additionally, the code can now be executed within a Docker installation, further increasing flexibility and stability. ### Batch summarization in message history To address the crashes caused by the summarization function, batch summarization has been introduced in the message history class. This enhancement ensures that the context length stays within the model's token limit, maintaining the integrity of the conversation history. ## CI/CD and Testing Several improvements and fixes have been made to the continuous integration/continuous deployment (CI/CD) pipeline and testing processes. These updates include caching Python packages in the CI pipeline, skipping flaky challenges, and fixing issues related to benchmarks, regression tests, and environment configuration. These improvements ensure a smoother development and testing experience for contributors. ## New Contributors We would like to extend a warm welcome to our new contributors: @DrMurx, @BorntraegerMarc, @javableu, @scenaristeur, and other active community members. Their valuable contributions helped shape this update and push the project forward. We look forward to their continued involvement in the Auto-GPT community. ## Conclusion In conclusion, the latest update to the Auto-GPT GitHub repository introduces several significant improvements, including directory-based plugins, keyboard navigation, JSON fixes, and more reliable Python code execution. The team has also made enhancements to the CI/CD pipeline and testing processes. We would like to extend our gratitude to all the contributors who made these changes possible. For a comprehensive list of all the changes made in this update, please refer to the [full changelog](https://github.com/Significant-Gravitas/Auto-GPT/compare/v0.4.0...v0.4.1). Stay tuned for more updates and new releases from the Auto-GPT project. Remember to join the discussion boards to connect with fellow developers and share your thoughts and ideas. Happy coding!
Last changed by  
 
Roberto Henríquez Perozo
72

Read more

evo-ninja

Welcome to evo.ninja.

8/19/2023
# Grego.AI B2C Strategy

In order to develop a comprehensive framework for the B2C strategy I recommend:

7/3/2023
AutoGPT Release v0.4.3

Introduction

6/28/2023
Untitled

Polywrap Newsletters: AI and LLMs Updates - June 2023 1. Introducing VOYAGER: An Open-Ended Embodied Agent with Large Language ModelsThe Minecraft-based AI, VOYAGER, leverages large language models to develop a set of growing skills and implement complex behaviors. Its robust learning abilities make it a standout project in the AI landscape. Read more about this project here. 2. Gorilla: Bridging the Gap between LLMs and APIsGorilla is an innovative solution teaching large language models how to accurately invoke APIs. With a growing store of APIs, Gorilla aims to enhance the capability of LLMs significantly. Find out more about this project here. 3. LLMs as Tool MakersDiscover how large language models can become tool makers with a closed-loop framework, enabling them to create their own reusable tools. It's an intriguing approach that offers cost-effective operations without sacrificing the quality of generated tools. Learn more about this project here. 4. ReWOO: Efficient Augmented Language Models through Reasoning-Observation DecouplingReWOO is a new paradigm that reduces computation complexity by separating the reasoning process of LLMs from external observations. With proven enhancements across several NLP benchmarks, ReWOO shows a promising future. Learn more about ReWOO here. And the Code Repository 5. FastChat's OpenAI-compatible API Server by lmsys.orgFastChat's OpenAI-compatible API server offers an open-source alternative for applications built on OpenAI APIs. This makes it possible to port applications to use open-source alternatives without modifying the code. Read more about this announcement on their blog post.

6/12/2023
Read more from Roberto Henríquez Perozo
Published on HackMD