Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/perpendicularai/sekernel_for_llm
This is a Python module used to create a semantic kernel in your OpenAI API compatible chat applications.
https://github.com/perpendicularai/sekernel_for_llm
chat generative-ai llamacpp python3 semantic-kernel
Last synced: 7 days ago
JSON representation
This is a Python module used to create a semantic kernel in your OpenAI API compatible chat applications.
- Host: GitHub
- URL: https://github.com/perpendicularai/sekernel_for_llm
- Owner: perpendicularai
- License: mit
- Created: 2024-07-17T11:21:04.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2024-07-31T04:02:50.000Z (2 months ago)
- Last Synced: 2024-09-22T19:02:05.250Z (11 days ago)
- Topics: chat, generative-ai, llamacpp, python3, semantic-kernel
- Language: Python
- Homepage:
- Size: 66.4 KB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# 🃏 SeKernel_for_LLM
This is a Python module used to create a semantic kernel in your openai api compatible chat applications.
### 🍬 Features
- In-chat memory
- Internet-search
- Database querying
## ⚙️ How to:
- Clone the repo and import the modules into your project. Ensure that it is in the project directory.
- ```python
import kernel
import plugins
### INTERNET-SEARCH ###
# Define search plugin
search_prompt = plugins.searchPlugin(output=question) # If context equals None, use the Chat template. See `kernel.py` for more templates.
# Initialize the kernel
data = kernel.shopTemplate(prompt=prompt, plugin=plugins.defaultPlugin(), context=search_prompt or context=None # Where no context is provided, and so you may assume the AI assistant to not have any awareness of information of events that took place after the date until which it's training data is up until) # See plugins.py module for more plugins
### DATABASE ###
# Using this database plugin
# Initialize the database plugin
db = plugins.dbConn()
# Use the database plugin along with the dbChatPlugin
data = kernel.chatTemplate(prompt=prompt, plugin=plugins.dbChatPlugin())
# Excuting the query
db.execute(response)
# Getting the output
response = db.fetchall()
### LlamaCpp ###
# Parsing the kernel model to LlamaCpp
LlamaCpp
client = Llama(
model_path=kernel.model() # Make sure to add your GGUF model in the kernel module.
)
# Use the kernel and set messages parameter equal to data. Depending on your LLM API defintion, messages may be a different parameter, in this case it is messages, as defined in the OpenAI API definition.
output = client.create_chat_completions(
messages = data
)
```
See [OpenAI](https://platform.openai.com/docs/api-reference/chat/create) API reference for more.
- ```python
# You may then append any new content and/or messages to the kernel
data.append(new_message)
```
## 📽️ Short Films
See examples of using the SeKernel_for_LLM with 🦙 [LlamaCpp Python bindings](https://github.com/abetlen/llama-cpp-python)
#### The square-root of 2
https://github.com/user-attachments/assets/cb48e962-2cba-4672-b4e7-c0f77455bb74
#### Internet search with Google for the price of leggings
https://github.com/user-attachments/assets/fdf5ac16-c8b7-4b39-b91b-d4e2d8d2d888
#### Database query
https://github.com/user-attachments/assets/ad9c87a4-475f-4ca0-a576-109709ca84b0
I hope this helps someone starting out. :octocat: Happy prompting!!!