first install all requirements:

pip3 install -r requirements.txt

Then, download the LLM model and place it in a model directory:

• LLM: default to ggml-gpt4all-j-v1.3-groovy.bin.

Note: because of the way langchain loads the SentenceTransformers embeddings, the first time it will requires internet connection to download the embeddings model itself.

Add files to source documents folder

The supported extensions are:

• .csv: CSV,
• .docx: Word Document,
• .doc: Word Document,
• .enex: EverNote,
• .eml: Email,
• .epub: EPub,
• .html: HTML File,
• .md: Markdown,
• .msg: Outlook Message,
• .odt: Open Document Text,
• .pdf: Portable Document Format (PDF),
• .pptx : PowerPoint Document,
• .ppt : PowerPoint Document,
• .txt: Text file (UTF-8),

Run the following command to ingest all the data.

python ingest.py

Note: during the ingest process no data leaves local environment.

Please put model=GPT4All in .env file In order to ask a question, run a command like:

python ampcome_gpt.py

And can chat with it.

Type exit to finish the script.

Please put model=Palm in .env file,

• added memory buffer.
• added some prompt engineering as well

In order to ask a question, run a command like:

python ampcome_gpt.py

And can chat with it.

Type exit to finish the script.