Skip to content

Commit

Permalink
Improved Messaging
Browse files Browse the repository at this point in the history
  • Loading branch information
@KillianLucas
KillianLucas committed Aug 23, 2023
1 parent 43ad3c4 commit a46f97e
Showing 1 changed file with 34 additions and 29 deletions.
63 changes: 34 additions & 29 deletions interpreter/interpreter.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@
from .llama_2 import get_llama_2_instance

import os
import time
import platform
import openai
import getpass
Expand All @@ -16,7 +17,7 @@
from rich import print
from rich.markdown import Markdown

# Function schema for GPT-4
# Function schema for gpt-4
function_schema = {
"name": "run_code",
"description":
Expand All @@ -42,7 +43,9 @@
# Message for when users don't have an OpenAI API key.
# `---` is at the bottom for aesthetic reasons.
missing_api_key_message = """
🔑 **OpenAI API key not found.**
---
🔑 OpenAI API key not found.
To use `GPT-4` (recommended) please provide an OpenAI API key. You can [get one here](https://platform.openai.com/account/api-keys).
Expand Down Expand Up @@ -81,9 +84,9 @@ def __init__(self):
# (blocks are visual representation of messages on the terminal)
self.active_block = None

# Note: While Open Interpreter can use Llama, we will prioritize GPT-4.
# GPT-4 is faster, smarter, can call functions, and is all-around easier to use.
# This makes GPT-4 better aligned with Open Interpreters priority to be easy to use.
# Note: While Open Interpreter can use Llama, we will prioritize gpt-4.
# gpt-4 is faster, smarter, can call functions, and is all-around easier to use.
# This makes gpt-4 better aligned with Open Interpreters priority to be easy to use.
self.llama_instance = None

def cli(self):
Expand Down Expand Up @@ -122,7 +125,7 @@ def get_info_for_system_message(self):

elif self.local:

# Tell Llama-2 how to run code.
# Tell llama-2 how to run code.
# (We actually don't use this because we overwrite the system message with a tiny, performant one.)
# (But someday, when Llama is fast enough, this should be how we handle it.)
info += "\n\nTo run Python code, simply write a Python code block (i.e ```python) in markdown. When you close it with ```, it will be run. You'll then be given its output."
Expand All @@ -140,15 +143,15 @@ def chat(self, message=None, return_messages=False):

# Connect to an LLM (an large language model)
if not self.local:
# GPT-4
# gpt-4
self.verify_api_key()

# ^ verify_api_key may set self.local to True, so we run this as an 'if', not 'elif':
if self.local:
# Llama-2
# llama-2
if self.llama_instance == None:

# Find or install LLama-2
# Find or install llama-2
self.llama_instance = get_llama_2_instance()

# If the user decided not to download it, exit gracefully
Expand All @@ -163,10 +166,10 @@ def chat(self, message=None, return_messages=False):

# If self.local, we actually don't use self.model
if not self.local and "gpt-3.5" in self.model:
welcome_message += f"\n> Model set to `{self.model}`."
welcome_message += f"\n> Model set to `{self.model.upper()}`."

if self.local:
welcome_message += f"\n> Model set to `llama-2`."
welcome_message += f"\n> Model set to `Llama-2`."

# If not auto_run, tell the user we'll ask permission to run code
# We also tell them here how to exit Open Interpreter
Expand Down Expand Up @@ -236,26 +239,28 @@ def verify_api_key(self):
if 'OPENAI_API_KEY' in os.environ:
self.api_key = os.environ['OPENAI_API_KEY']
else:
# Print message with newlines on either side (aesthetic choice)
print('', Markdown(missing_api_key_message), '')
# This is probably their first time here!
print('', Markdown("# **Welcome to Open Interpreter.**"), '')
time.sleep(1)

print(Markdown(missing_api_key_message), '')
response = input("OpenAI API key: ")
import time

if response == "":
# User pressed `enter`, requesting Llama-2
# User pressed `enter`, requesting llama-2
self.local = True

print(Markdown("> Switching to `llama-2`...\n\n**Tip:** Run `interpreter --local` to automatically use `Llama-2`."), '')
time.sleep(3)
print(Markdown("---"))
print(Markdown("> Switching to `Llama-2`...\n\n**Tip:** Run `interpreter --local` to automatically use `Llama-2`."), '')
time.sleep(2)
print(Markdown("---"), '')
return

else:
self.api_key = response
print(Markdown("> Model set to `gpt-4`"))
print(Markdown("> Model set to `GPT-4`"))
# time.sleep(1)
print('', Markdown("**Tip:** To save this key for later, run `export OPENAI_API_KEY=your_api_key` on Mac/Linux or `setx OPENAI_API_KEY your_api_key` on Windows."), '')
time.sleep(3)
time.sleep(2)
print(Markdown("---"))

openai.api_key = self.api_key
Expand All @@ -272,12 +277,12 @@ def respond(self):
system_message = self.system_message + "\n\n" + info

if self.local:
# While Llama-2 is still so slow, we need to
# While llama-2 is still so slow, we need to
# overwrite the system message with a tiny, performant one.
system_message = "You are an AI that executes Python code. Use ```python to run it."

# Model determines how much we'll trim the messages list to get it under the context limit
# So for Llama-2, we'll use "gpt-3.5-turbo" which (i think?) has the same context window as Llama-2
# So for llama-2, we'll use "gpt-3.5-turbo" which (i think?) has the same context window as llama-2
self.model = "gpt-3.5-turbo"

messages = tt.trim(self.messages, self.model, system_message=system_message)
Expand All @@ -289,7 +294,7 @@ def respond(self):

# Make LLM call
if not self.local:
# GPT-4
# gpt-4
response = openai.ChatCompletion.create(
model=self.model,
messages=messages,
Expand All @@ -298,7 +303,7 @@ def respond(self):
temperature=self.temperature,
)
elif self.local:
# Llama-2
# llama-2

# Turn function messages -> system messages for llama compatability
messages = self.messages
Expand Down Expand Up @@ -329,7 +334,7 @@ def respond(self):
if not self.local:
condition = "function_call" in self.messages[-1]
elif self.local:
# Since Llama-2 can't call functions, we just check if we're in a code block.
# Since llama-2 can't call functions, we just check if we're in a code block.
# This simply returns true if the number of "```" in the message is odd.
if "content" in self.messages[-1]:
condition = self.messages[-1]["content"].count("```") % 2 == 1
Expand Down Expand Up @@ -361,7 +366,7 @@ def respond(self):
# Now let's parse the function's arguments:

if not self.local:
# GPT-4
# gpt-4
# Parse arguments and save to parsed_arguments, under function_call
if "arguments" in self.messages[-1]["function_call"]:
arguments = self.messages[-1]["function_call"]["arguments"]
Expand All @@ -372,13 +377,13 @@ def respond(self):
"parsed_arguments"] = new_parsed_arguments

elif self.local:
# Llama-2
# llama-2
# Get contents of current code block and save to parsed_arguments, under function_call
if "content" in self.messages[-1]:
current_code_block = self.messages[-1]["content"].split("```python")[-1]
arguments = {"language": "python", "code": current_code_block}

# Llama-2 won't make a "function_call" property for us to store this under, so:
# llama-2 won't make a "function_call" property for us to store this under, so:
if "function_call" not in self.messages[-1]:
self.messages[-1]["function_call"] = {}

Expand All @@ -391,7 +396,7 @@ def respond(self):
if in_function_call == True:

if self.local:
# This is the same as when GPT-4 gives finish_reason as function_call.
# This is the same as when gpt-4 gives finish_reason as function_call.
# We have just finished a code block, so now we should run it.
llama_function_call_finished = True

Expand Down

0 comments on commit a46f97e

Please sign in to comment.