API
Create new secret key and install the library. Set environment variable permanently (API secret key).
pin install openai
# platform.openai.com/account/api-keys
sudo gedit ~/.bashrc
# export OPENAI_API_KEY=xxx
echo $OPENAI_API_KEY
Request
Extract the content of the assistant's message from the JSON response.
import openai
import os
# Setup OpenAI API key
openai.api_key = os.environ.get("OPENAI_API_KEY")
question = "What is flask python?"
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": question}],
max_tokens=256,
n=1,
stop=None,
temperature=0.7
)
answer = response['choices'][0]['message']['content']
print(answer)
"""
Flask is a lightweight web framework written in Python.
It is designed to be simple, easy to use, and flexible ...
"""
Streaming Completion
Streaming with the OpenAI API allows you to get partial results and process them as they become available, which can be more efficient and responsive.
import openai
import os, sys
# Setup OpenAI API key
openai.api_key = os.environ.get("OPENAI_API_KEY")
# User question
question = "What is flask python?"
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=[{"role": "user", "content": question}],
max_tokens=256,
n=1,
stop=None,
temperature=0.7,
stream=True
)
for chunk in response:
content = chunk["choices"][0]["delta"].get("content", "")
print(content, end="", flush=True)
Conversation History
You can create a conversation where the model remembers the context. You'll need to include that conversation history in subsequent requests.
import openai
import os, sys
# Setup OpenAI API key
openai.api_key = os.environ.get("OPENAI_API_KEY")
# Initialize conversation history
conversation_history = []
while True:
# User question
print("\nPlease enter your question (or 'exit' to end):")
question = input()
if question.lower() == 'exit':
break
# Add user question to conversation history
conversation_history.append({"role": "user", "content": question})
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=conversation_history,
max_tokens=256,
n=1,
stop=None,
temperature=0.7,
stream=True
)
for chunk in response:
content = chunk["choices"][0]["delta"].get("content", "")
print(content, end="", flush=True)
# Add API response to converstion history
conversation_history.append({"role": "system", "content": content})
print("Conversation ended.")
Context
You can add a context message at the start of the conversation to instruct the model.
import openai
import os, sys
# Setup OpenAI API key
openai.api_key = os.environ.get("OPENAI_API_KEY")
# Initialize conversation history
conversation_history = []
# Context for keeping answers short
context_message = {
"role": "system",
"content": "System: Please keep the answers short"
}
conversation_history.append(context_message)
# Question stream
questions = [
"What is Flask? Keep the answers short.",
"What's the current version?"
]
for question in questions:
print("\nQuestion:", question)
# Add user question to conversation history
conversation_history.append({"role": "user", "content": question})
response = openai.ChatCompletion.create(
model="gpt-3.5-turbo",
messages=conversation_history,
max_tokens=64,
n=1,
stop=None,
temperature=0.7,
stream=True
)
for chunk in response:
content = chunk["choices"][0]["delta"].get("content", "")
print(content, end="", flush=True)
# Add API response to converstion history
conversation_history.append({"role": "system", "content": content})
Last update: 291 days ago
