GO-CREW: How to Use Guide

Go back

Easy Steps to Get Started

Install Ollama: Visit ollama.com and follow the installation instructions for your operating system.
Install a model (like Llama3.1):
```
ollama pull llama3.1
```
If needed, edit the default.toml configuration file for GO-CREW.
Run GO-CREW:
```
./go-crew
```

Understanding the Configuration (default.toml)

name = 'default.toml'
prompt = 'Hello World'  # initial prompt
default_model = 'llama3.1'
default_temperature = 0.8
default_model_context = 0  # 0 = automatic
default_timeout = 60
# default_custom_output = '(.*)<\/Output>'   # example for marco-o1
concurrency = 2  # for more speed
language = ''    # for force a response language
system_prompt = ''
server_url = ''  # for custom servers
verbose = false

[hosting]
enabled = false  # host a webpage
port = 3000
auto_open = false  # open automatically when start

[embedding]
enabled = false         # use RAG functions
database = 'default.db'
database_compressed = false
database_encryption_key = ''
model = 'nomic-embed-text'  # model for embedding
contexts_rerank = true
contexts_per_request = 3

[embedding.chunking]
semantic_chunking = true    # use semantic chunking
semantic_threshold = 0.456
chunk_max_size = 1024
chunk_overlap = 154         # 0.15*chunk

[[embedding.sources]]
document = 'document.txt'  # read a file, can use * for name

[[embedding.sources]]
document = 'document.csv'   # read a csv file
header_question='question'  # optional, if you want to use a header for the question embedding
header_answer='answer'      # optional, if you want to use a header for the answer response

[[embedding.sources]]
url = 'https://www.jonathanhecl.com/go-crew/'   # read a website

Configuration Explanation

prompt: The initial prompt sent to the model.
default_model: The default language model to use (e.g., 'llama3.1').
default_temperature: Controls the randomness of the output (0.0 to 1.0).
default_model_context: Context size (0 for automatic).
default_custom_output: Regular expression for custom output.
concurrency: Number of concurrent operations for increased speed.
hosting: Settings for hosting a web interface.
embedding: Settings for RAG (Retrieval-Augmented Generation) functions.
embedding.sources: Define sources for document ingestion (files or URLs).