https://github.com/mya-mya/chatmpc

Natural Language-driven Controller Personalization Framework 📣
https://github.com/mya-mya/chatmpc

control-engineering natural-language-processing

Last synced: 29 days ago
JSON representation

Natural Language-driven Controller Personalization Framework 📣

• Host: GitHub
• URL: https://github.com/mya-mya/chatmpc
• Owner: Mya-Mya
• License: mit
• Created: 2023-06-26T02:29:30.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-02-28T05:35:57.000Z (10 months ago)
• Last Synced: 2024-02-28T06:33:09.512Z (10 months ago)
• Topics: control-engineering, natural-language-processing
• Language: MATLAB
• Homepage:
• Size: 20.5 KB
• Stars: 2
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
## ChatMPC

## Reproduce the Numerical Experiment

### 1. Download Sentence BERT Pre-trained Weights

Sentence BERT model is used in the intent extractor $f_\mathrm{int}$ in the interpreter $\mathcal{A}$. In the numerical experiment, we used `deepset/sentence_bert` as the pre-trained weights. 

Clone it from the [HuggingFace Model page](https://huggingface.co/deepset/sentence_bert).

### 2. Launch the IntentExtractor Python Server

Most of ChatMPC is implemented in MATLAB, but only the intent extractor $f_\mathrm{int}$ is implemented in Python. MATLAB code and the Python code communicate via HTTP.

Recall the path to the Sentence BERT folder, and execute the following:

```bash

>> cd IntentExtractor

>> python3 main.py --sbert_path 

```

Check that the server is running on `http://127.0.0.1:3838`.

### 3. Work in MATLAB

The code for the numerical experiment is `main.m`. 

Please run it **SECTION BY SECTION** because I'll need you to type/run some commands yourself in the middle of running.