Spaces:
Runtime error
Runtime error
File size: 4,509 Bytes
220a7fc d40a83a 220a7fc d40a83a 220a7fc 4097737 220a7fc a5f4443 cb4043a 6824b31 220a7fc d40a83a 220a7fc a5f4443 e9d8ee5 a5f4443 e9d8ee5 a5f4443 e09355d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 |
import io
import os
from openai import OpenAI
from langchain.tools import StructuredTool, Tool
from io import BytesIO
import requests
import json
from io import BytesIO
import chainlit as cl
def get_image_name():
"""
We need to keep track of images we generate, so we can reference them later
and display them correctly to our users.
"""
image_count = cl.user_session.get("image_count")
if image_count is None:
image_count = 0
else:
image_count += 1
cl.user_session.set("image_count", image_count)
return f"image-{image_count}"
def _generate_image(prompt: str):
"""
This function is used to generate an image from a text prompt using
DALL-E 3.
We use the OpenAI API to generate the image, and then store it in our
user session so we can reference it later.
"""
client = OpenAI()
response = client.images.generate(
model="dall-e-3",
prompt=prompt,
size="1024x1024",
quality="standard",
n=1,
)
image_payload = requests.get(response.data[0].url, stream=True)
image_bytes = BytesIO(image_payload.content)
print(type(image_bytes))
name = get_image_name()
cl.user_session.set(name, image_bytes.getvalue())
cl.user_session.set("generated_image", name)
return name
def _youtube_rag(prompt: str):
openai.api_key = os.environ["OPENAI_API_KEY"]
flattened_texts = []
#check if db exists
if os.path.exists(persist_directory):
#don't process transcripts
if debug:
print("Database exists, skipping transcript processing...")
else:
print("Database does not exist")
if debug:
print("Initializing database...")
docsearch = initialize_chroma_db(flattened_texts)
docs = docsearch.get_relevant_documents(prompt)
chat_model = ChatOpenAI(model_name="gpt-4-1106-preview")
chain = load_qa_chain(llm=chat_model, chain_type="stuff")
answer = chain.run(input_documents=docs, question=prompt)
return answer
def generate_image(prompt: str):
image_name = _generate_image(prompt)
return f"Here is {image_name}."
# this is our tool - which is what allows our agent to generate images in the first place!
# the `description` field is of utmost imporance as it is what the LLM "brain" uses to determine
# which tool to use for a given input.
generate_image_format = '{{"prompt": "prompt"}}'
generate_image_tool = Tool.from_function(
func=generate_image,
name="GenerateImage",
description=f"Useful to create an image from a text prompt. Input should be a single string strictly in the following JSON format: {generate_image_format}",
return_direct=True,
)
def youtube_rag(prompt: str):
answer = _youtube_rag(prompt)
return f" {answer}."
import os
import openai
from langchain.chat_models import ChatOpenAI
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.chains.question_answering import load_qa_chain
from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings.sentence_transformer import SentenceTransformerEmbeddings
debug = False
persist_directory = 'db'
embedding_function = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
# Function to initialize or load the Chroma database
def initialize_chroma_db(texts):
if os.path.exists(persist_directory):
# Load existing database
if debug:
print("Loading existing database...")
db = Chroma(persist_directory="./db", embedding_function=embedding_function)
else:
# Create and initialize new database
#embeddings = OpenAIEmbeddings()
if debug:
print("Creating new database...")
db = Chroma.from_texts(texts, embedding_function, persist_directory=persist_directory)
return db.as_retriever()
# thisis the youtube rag tool - which is what allows our agent to rag the youtube vector db
# the `description` field is of utmost importance as it is what the LLM "brain" uses to determine
# which tool to use for a given input.
youtube_rag_format = '{{"prompt": "prompt"}}'
generate_image_tool = Tool.from_function(
func=youtube_rag,
name="Youtube_Rag",
description=f"Useful to query the vector database containing youtube transcripts about Aaron Lebauer. Input should be a single string strictly in the following JSON format: {youtube_rag_format}",
return_direct=True,
)
|