SkyChat Chinese Chatbot GPT3

SkyChat

Simplified Chinese

English

Model introduction

SkyChat is a chat robot project developed based on the Chinese GPT-3 API. In addition to basic chat, dialogue, and question-and-answer, it can also support Chinese-English translation, content continuation, couplets, writing ancient poems, and generating recipes. , third-person narration, creating interview questions and other functions.

For more details, please see the official website documentation of Singularity Intelligence

Here are some examples:

Effect example

For experience and trial, please visit Singularity Intelligence Source API Trial

chat

Q&A

Generate recipes

enter:

Output:

pair of couplets

————————————————————————————

Demo usage tutorial:

Model service and Python environment construction

Windows

1. Download Anaconda and check the Add to environment variables option
   
2. Download and unzip the semantic_score_clean and user_profile_clean services, models, and sample training data
3. Open Anaconda and enter
   conda create -n semantic creates a new environment
   conda info --envs view environment
   activate semantic startup environment
   cd C:你的路径semantic_score_cleansemantic_score_clean cd to the unzipped semantic_score_clean folder
   python -m pip uninstall numpy deletes the initial numpy
   python -m pip install -r requirements.txt or pip install -r requirements.txt to install the required dependency packages
   python semantic_score_api.py runs the service
4. Download Cuda corresponding to the system and install it
5. Open a new Anaconda interface and enter
   conda create -n userprofile creates a new environment
   conda info --envs view environment
   activate userprofile start environment
   cd C:你的路径user_profile_cleanuser_profile_clean cd to the unzipped user_profile_clean folder
   python -m pip uninstall numpy deletes the initial numpy
   python -m pip install -r requirements.txt or pip install -r requirements.txt to install the required dependency packages
   python -m pip install torch==1.11.0+cu115 -f https://download.pytorch.org/whl/torch_stable.html download cuda version torch
   python server_v3.py runs the service

---

Unity version Demo

1. Download the latest package in Release.
2. Click Register on the openAPI website or demo to register and authenticate to obtain api-key and api-secret
3. Click Settings to enter the setting interface. Fill in the key and secret in the corresponding positions, as well as the names of both parties. Click Save and Exit after completion.
4. Confirm the semantic_score_clean and user_profile_clean services to be deployed locally. If deployed remotely, please replace the corresponding request URL in the settings.
5. Let's chat

---

Python version Demo

1. Click Register on the openAPI website or demo to register and authenticate to obtain api-key and api-secret
2. Open 你的路径OpenAPIDemoPythonDemomain.py fill in api-key and api-secret in the corresponding locations and modify the parameters that need to be modified.
3. Open Anaconda and enter
   conda create -n talk creates a new environment
   conda info --envs view environment
   activate talk startup environment
   cd C:你的路径OpenAPIDemoPythonDemo navigate to the folder where the code is located
   python -m pip install requests download dependent resources
   python main.py starts chatting (need to confirm 2 services to start)

---

*Minimum configuration required: Nvidia Gfx 1060

Join the developer group

If you have any questions, you may wish to scan the QR code on WeChat to join the developer group——

Don’t forget to star if you are interested~