v9.1.0-DEV

New

• Clear queue time.
• Added some notes to the pre-defined LLaMA-CPP models.
• New pre-defined models for StableDiffusion-CPP.

Updated / Fixed

• Fixed some queue errors.
• Replaced commands.
• Updated internet search.
• Updated default system messages.

Working on

• Video Pre-Training implementation.
• Support for other file types like PDF, CSV, etc. (only for the chatbots!).

Compatibility

All the servers and clients from this version has full compatibility with servers and clients from this same version and versions: v9.0.0.

v9.0.0

New

• Predefined models for LCPP chatbots.
• Use custom dtype for Transformers.
• RVC now allows to configure the number of threads to use (experimental).
• The server cache files are now cleared when the server closes or every 5 minutes.
• Internet search can now search for news and websites.
• The clients can now check if a chatbot is multimodal or not.
• Simulated vision, for chatbots that are not multimodal.
• Python client bindings (experimental, not tested).

Fixed / Updated

• Data share can be turned on and off by both the server and user.
• Internet search now uses DuckDuckGo instead of google.
• Bug fixes.
• Updated system prompts.
• Updated requirements.
• Updated API keys, they can have uppercase, numbers and symbols now.
• Changed the license from TAO71 License 111 to TAO71 I4.0 License (version 1).

Compatibility

All the servers and clients from this version has full compatibility with servers and clients from this same version.

v8.2.0

New

• Multimodal chatbot.
• Re-programmed the conversations (old conversations are compatible!).
• New config.

Fixed / Updated

• Updated the requirements.
• Updated documentation.
• Fixed a bug in the C# PyAIResponse.cs file.

Working on

• Bug fixes in the server.
• Python client bindings.

Note

This version clients are compatible with the servers from the versions: v8.0.0, v8.1.0-DEV.
Also, new version servers are compatible with the clients from the mentioned versions.

This changes are from the DEV versions:

• v8.1.0-DEV

v8.1.0 DEV

New

• You can now use StableDiffusion-CPP-Python for text2img.
• Custom system prompts based in if the API key is admin or not.
• The user can now select if it wants to enable or disable some specific system prompts of the server.
• New development branch.

Fixed / Updated

• Fixed a bug with the queues.
• Probably fixed a bug that causes the server to not delete the temporal files.
• Fixed a bug with the data share.
• Updated the documentation to this new version.
• Fixed a glitch that alters system prompts to add server's system prompts, even if the user said no.
• Updated default system prompts.
• Added installation for StableDiffusion-CPP-Python.
• Updated requirements.
• Fixed a bug related with the API keys.
• Updated C# client.
• Updated TypeScript client.
• Deprecated the test.py script.
• Added the verbose = True argument in the Speech Recognition.
• Deleted plugins for now.

Working on

• Multimodal chatbot.

Note

This new version clients are compatible with the servers from the versions: v8.0.0.
Also, new version servers are compatible with the clients from the mentioned versions.

Warning

Remember that this is a development version, bugs, crashes and incompleted code is expected!

v8.0.0

New

• Moved assets to this repository (I4.0's assets repository is now deprecated).
• New and updated documentation.
• Splitted different chatbots across 3 files.
• New configuration.
• I4.0 can now store memories across conversatios.
• New server functionality.
• More security in the servers.
• Added some code that deletes the generated files from the server when processing is done.
• Files now have a maximum size allowed by the server.
• New system prompts.
• Support for Intel GPUs in most of services.

Fixed / Updated

• Fixed and updated audio generation.
• Fixed speech recognition using hf.
• Fixed depth estimation.
• Fixed image to text.
• Fixed and updated image to image.
• Updated some functionalities of the chatbots.
• Fixed some bugs in Python 3.11.
• When deleting a conversation it now gets fully deleted, instead of replacing it with empty strings.
• Fixed some server errors that made the server crash.
• Clients can now send files to the server, instead to sending it to another server.
• Fixed some other bugs in the server.
• Fixed some bugs related to the conversations.
• Update conversation templates.
• Updated the automatic installation of requirements.
• Updated requirements.
• Updated test.
• Updated both C# and TypeScript bindings.
• Updated the readme with the latest information.

Working on

• Support for Intel GPUs in RVC.
• Multimodal chatbot.
• TextToImage generation with StableDiffusion.CPP.

Note

This new version clients are compatible with the servers from this same version (v8.0.0).
Also, new version servers are compatible with the clients for this same version.

v7.1.0

Added:

• Python test for server.
• TypeScript client bindings (NOT TESTED).
• System sound recorder.
• You now have a function to get the Terms Of Service from the server.
• Server checks if you have a TOS.txt file. If you don't, the server creates it.
• User now can get a conversation from the server.
• User now can get the Terms of Service from the server.
• User now can get information about it's API key from the server.
• User now can get a description of a service.
• New service: ld (Language Detection).
• API keys can now be admin of the server. This can be used to execute some commands on the server.

Changed:

• Record microphone and system sound is now programmed in C#.
• New configuration that allows you to have multiple models of the same type. For more information and how to use please read the file LibI4/Python_AI/ai_config.py.
• Users can choose which model they want to use.
• The server now uses collections.abc.
• Server now have no send and receive bytes limit.

Deprecated:

• Server for receive files is now deprecated. Since the server now has no bytes limit, there's no reason to receive files using this server instead of the default server.
• UVR service is deprecated temporally, but it will be re-programmed and will be working again.

Working on:

• Official client.

TODO:

• New Discord bot.
• Compatibility with Intel GPUs.
• More languages support for the client (Python).
• Better Docker support.
• Documentation.

Warning

This version has not been competelly tested. Errors and bugs are expected.

Note

This new version clients are compatible with the servers from the versions: v7.0.0, v7.0.1.
Also, new version servers are compatible with the clients from the mentioned versions.
Next version will focus on the chatbot, bug fixes and more testing.

I4.0 1st Birthday!!!

Note

This is not an update, it's just a message for the community.

Today we celebrate I4.0's first birthday! 🎉

Some time ago, we set out with a clear mission: to create an artificial intelligence as close to AGI as possible—free of limitations, open-source, and available to everyone.

This milestone isn’t just for I4.0, but for everyone who believes in innovation without barriers, in a future where AI remains an open-source and accessible tool for all.

Let’s keep moving forward together towards a future without limits!

We’re currently working on optimizations, new modules and client bindings in Python, and an update will be released soon.

Additionally, I4.0's official server (tao71.sytes.net) now offers some free services! While these options may change over time, here’s what’s currently available for free:

• A free basic chatbot (at index 0)
• Text Classification (at index 0)
• NSFW filters for both text and images (at index 0 for both)

v7.0.1

Changed:

• Fixed a bug of the chatbot when using GPT4All.
• Updated requirements.

v7.0.0

Deprecated:

• TCFG to JSON.
• Intel GPUs are not supported anymore (for now). This is because we can't test I4.0 on Intel GPUs.

Removed:

• Thinking prompts, because it only consumed resources and don't work well in some models.
• Some dependencies from C#.
• Some unnecesary data that the server sends to the client.

Changed:

• Added some more requirements.
• Banning, now it can ban IPs and keys.
• Internet connection, reading from websites.
• Auto-install requirements, you can now force to download the CPU wheel of PyTorch and LLAMA-CPP-Python.
• Data share banning.
• System prompts, replaced with more and better system prompts.
• whisper service is now called speech2text, because in a future it will support more speech2text models than whisper.
• Inference services are now separated by directories.
• Configuration, replaced some options with others, added a few more and deleted unnecesary/deprecated options.
• Print loading message is not optional now.
• License, changed some conditions and added a few more.
• Fixed some bugs.
• Some other things.

Added:

• New chatbot system available: LLAMA-CPP-Python (Python bindings of LLAMA-CPP).
• Docker support. WARNING: Only tested on AMD GPUs. The support it's kinda bad for now, but it will get better in the future.
• Some computers where we test I4.0 on the README.md file.
• New re-programmed server and client that allows streaming and more things! WARNING: Older versions servers WILL NOT be compatible with new versions clients and vice versa.
• Faster client.
• Better chatbot.
• Add better support for some models of some services.
• Some other things.

Working on:

• Documentation for this version.
• Official C# client.
• New discord bot.

TODO:

• System sound recorder (required by the Video Reaction plugin).
• Better Docker support.
• Compatibility with Intel GPUs.
• More languages support for the client (Python and TypeScript).

Warning

C# plugins HAS NOT been tested with the new client.
This was ONLY TESTED using Arch Linux OSs.

v6.5.0 BETA-2

• Bug fixes.
• Removed compatibility with StyleTTS2 (for now) due to it's requirements conflicting with I4.0's requirements.
• New System Prompts.
• New prompt template that allows I4.0 to think.
• New configuration.
• Updated requirements.
• Discord bot it's still bugged/broken.
• So much more!

Warning

THIS IS A BETA RELEASE, BUGS ARE EXPECTED.
THE FINAL 6.5.0 RELEASE ALSO MIGHT BE DIFFERENT.

Note

If you find any bug, it will help us if you describe it in the issues tab.
Also, next release might be the full v6.5.0!