Chat AI: A chatbot framework
Project description
Chat AI: Run LLMs with your own hardware
Chat AI is an interface that sits on top of a custom engine which is responsible for processing requests and responses to and from an active Google T5-Flan model.
Bundled Installation
Official Build can be acquired here for those who want to use a compiled version of Chat AI without having to install any additional requirements
Development
Prerequisites
- Ubuntu 20.04+ or Windows 10+
- Python 3.10.6
- pip-23.0.1
Pypi installation
Use this installation method if you intend to use Chat AI from the command line or with other python libraries or scripts.
pip install chatairunner
pip install git+https://github.com/w4ffl35/diffusers.git@ckpt_fix
pip install git+https://github.com/w4ffl35/transformers.git@tensor_fix
Development installation
Use this installation method if you intend to modify the source code of Chat AI.
- Ubuntu 20.04+ or Windows 10+
- Python 3.10.6
- pip-23.0.1
- Fork this repo on github
git clone https://github.com/Capsize-Games/chatai
python -m venv env
- skip if you don't want to use venvcd chatai && pip install -e .
cd chatai && python main.py
Here are the special tokens used by the model:
- : Used to pad sequences to a fixed length.
- : Beginning of sequence token, used to indicate the start of a sequence.
- : End of sequence token, used to indicate the end of a sequence.
- : Unknown token, used to represent out-of-vocabulary words.
- : Mask token, used for masked language modeling tasks.
- <extra_id_X>: Special token used to represent additional task-specific labels, where X is a number from 0 to 99.
- : End of turn token, used in conversation models to indicate the end of a speaker's turn.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
chatairunner-1.0.7.tar.gz
(27.0 kB
view hashes)
Built Distribution
Close
Hashes for chatairunner-1.0.7-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 376ea475cf636d69dab5fb88d50788589a17ae640ade9c325c738447b41d1fa1 |
|
MD5 | 2ae52cbbc52e9e5f636c8461ad71427b |
|
BLAKE2b-256 | d97e52abdde280d12ee86259b74635685ae512e0a20e97496ccedbe7f185af6d |