๐ฅ๐ฅLatest Release Version:V0.0.1
๐ฅ๐ฅMulti-Models Management
Using large language models and multi-agent technology, a single request can automatically generate marketing copy, images, and videos, and with one click, can be sent to multiple platforms, achieving a rapid transformation in marketing operations.
| LLM | Supported | Model Type | Notes |
|---|---|---|---|
| ChatGPT | โ | Proxy | Default |
| Bard | โ | Proxy | |
| Vicuna-13b | โ | Local Model | |
| Vicuna-13b-v1.5 | โ | Local Model | |
| Vicuna-7b | โ | Local Model | |
| Vicuna-7b-v1.5 | โ | Local Model | |
| ChatGLM-6B | โ | Local Model | |
| ChatGLM2-6B | โ | Local Model | |
| baichuan-13b | โ | Local Model | |
| baichuan2-13b | โ | Local Model | |
| baichuan-7b | โ | Local Model | |
| baichuan2-7b | โ | Local Model | |
| Qwen-7b-Chat | Coming soon | Local Model |
| LLM | Supported | Notes |
|---|---|---|
| sentence-transformers | โ | Default |
| text2vec-large-chinese | โ | |
| m3e-large | โ | |
| bge-large-en | โ | |
| bge-large-zh | โ |
Firstly, download and install the relevant LLMs.
mkdir models & cd models
# Size: 522 MB
git lfs install
git clone https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2
# [Optional]
# Size: 94 GB, supported run in cpu model(RAM>14 GB). stablediffusion-proxy service is recommended, https://github.com/xuyuan23/stablediffusion-proxy
git lfs install
git clone https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
# [Optional]
# Size: 16 GB, supported run in cpu model(RAM>16 GB). Text2Video service is recommended. https://github.com/xuyuan23/Text2Video
git lfs install
git clone https://huggingface.co/cerspense/zeroscope_v2_576w
Then, download dependencies and launch your project.
yum install gcc-c++
pip install -r requirements.txt
# copy file `.env.template` to new file `.env`, and modify the params in `.env`.
cp .env.template .env
[Options]
# deploy stablediffusion service, if StableDiffusion proxy is used, no need to execute it!
python operategpt/providers/stablediffusion.py
[Options]
# deploy Text2Video service, if Text2Video proxy server is used, no need to execute it!
python operategpt/providers/text2video.py
# Quick trial: two params: idea and language, `en is default`, also supported zh(chinese).
python main.py "Prepare a travel plan to Australia" "en"
- By default, ChatGPT is used as the LLM, and you need to configure the
OPEN_AI_KEYin.env
OPEN_AI_KEY=sk-xxx
# If you don't deploy stable diffusion service, no image will be generated.
SD_PROXY_URL=127.0.0.1:7860
# If you don't deploy Text2Video service, no videos will be generated.
T2V_PROXY_URL=127.0.0.1:7861- More Details see file
.env.template