hayloft 0.4.1

UI tool for LLM frameworks

Hayloft

UI tool for LLM frameworks to make easy prompt/completion tracking, store and comparison of different sessions.

• Installation
• Usage
  • LlamaIndex
  • BabyAGI

Installation

Install package with pip

pip install hayloft

Usage

Start hayloft server

hayloft start

Trace logs of your script on http://localhost:7000

LlamaIndex

Install llama-index, create example.py file as below. Put examples folder from llama_index repo near the file.

import os
os.environ["OPENAI_API_KEY"] = 'YOUR_OPENAI_API_KEY'

from llama_index import SimpleDirectoryReader, VectorStoreIndex 
from hayloft.llama_index import grab_logs

grab_logs()

documents = SimpleDirectoryReader("examples/paul_graham_essay/data").load_data()
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()
query_engine.query("What did the author do growing up?")

Or you can start live sessions from hayloft ui, just modify code like here

import os
os.environ["OPENAI_API_KEY"] = 'YOUR_OPENAI_API_KEY'

from llama_index import SimpleDirectoryReader, VectorStoreIndex 
from hayloft.llama_index import listen

def agent(query: str):
    documents = SimpleDirectoryReader("examples/paul_graham_essay/data").load_data()
    index = VectorStoreIndex.from_documents(documents)
    query_engine = index.as_query_engine()
    query_engine.query(query)

listen(agent)

Start this script

python example.py

BabyAGI

Clone BabyAGI fork repo, setup virtual env and install all dependencies

git clone [e-mail]:eturchenkov/babyagi-hayloft.git && cd babyagi-hayloft
pip install -r requirements.txt

Adjust config in .env file and start babyagi.py script

python babyagi.py