sesgx-cli

CLI to perform experiments with the SeSG framework.

Usage

Development

Create a virtual environment:

python -m venv .venv

Activate the virtual environment:

source .venv/bin/activate  # if using linux

Install the project in editable mode:

pip install -e .

To install the optional dependencies groups, use the following command:

pip install -e ".[group-name]"

For example, if you want to install lda-topic-extraction and bert-word-enrichment, run the following command:

pip install -e ".[lda-topic-extraction,bert-word-enrichment]"

---

Run instructions

Using Ollama for llm strategy

To run llm models using Ollama please install it based on these instructions:

• Ollama Download

Once ollama is properly installed you can download a model using:

ollama pull {model_name}

All models available here.

More commands can be discovered using:

ollama -h

Be aware that each model will require a minimum of hardware specs.

Using OpenAi models for llm strategy

To use any OpenAi models it is necessary to provide a key. This is sensible information please use .env, do not expose private keys.

Enter your key using "OPENAI_API_KEY" value. See .env.example.

Be aware that OpenAi api is a paid service.

---

Telegram report

The sesg experiment start and sesg scopus search commands has a optional parameter name telegram_report (-tr). This option enable a execution report to be sent via telegram.

1. Follow these instructions to set a telegram bot.

2. Create a group and add the bot created.

   Tip: make sure tour bot has the correct admin rights. Through @BotFather and set Manage Topics to be true. And also allow it to be added in groups.

3. Enable the topic feature of the group.

   Tip: send a message to your newly created group, go to group info, under reactions there will be a toggle switch for the topics option.

4. You can get the group chat id using this request directly in your browser:

      https://api.telegram.org/bot<YourBOTToken>/getUpdates

5. Get its token and your chat id. Add these variable to your .env file.

      TELEGRAM_TOKEN="" # bot token.
      TELEGRAM_CHAT_ID_EXPERIMENT="" # your chat id for the experiment chat.
      TELEGRAM_CHAT_ID_SCOPUS="" # your chat id for the scopus chat.
      PC_SPECS="" # your hardware specs.

⚠️ Note: this works as a local bot. To only store your experiments information, such as time of execution or execution checkpoints. This is not a live server conversational bot.