vanildo/rag-handson

RAG HandsOn

Pré-requisitos

• Docker daemon/cli client
• Miniconda
• Git
• VSCode

Conda

1. Criar ambiente virtual

conda create --name ml -c conda-forge python=3.11
conda activate ml

2. Instalar dependências

conda install -c conda-forge jupyter pandas numpy sentence-transformers tensorflow 
pip install -U ibm-generative-ai "ibm-generative-ai[langchain]" pypdf readchar weaviate-client python-dotenv langchain huggingface torch gradio chromadb

3. Testar dependências

curl -LJO https://github.com/vanildo/rag-handson/raw/main/configTest.py
python ./configTest.py

Elasticsearch

• https://www.elastic.co/guide/en/elasticsearch/reference/current/date.html
• Baixar o arquivo docker-compose.yml

curl -LJO https://github.com/vanildo/rag-handson/raw/main/docker-compose.yml

Rodar Elasticsearch local

docker compose up -d

Weaviate (Opcional - alternativa ao ElasticSearch)

• Site: https://weaviate.io
• Github: https://github.com/weaviate/weaviate
• Baixar o arquivo docker-compose.yml

curl -LJO https://github.com/vanildo/rag-handson/raw/main/docker-compose-weaviate.yml

para rodar o docker compose com suporte a persistência, tem que cria a pasta weaviate_data.

mkdir weaviate_data

• Comando para rodar local:

docker compose -f ./docker-compose-weaviate.yml up -d