Updated litellm and openai

[Notnaton]

Describe the changes you have made:

Updated LiteLLM and OpenAI
Needed to add a check if content is None.
Only tested with LM Studio

Reference any relevant issue (Fixes #000)

#765

• I have performed a self-review of my code:

I have tested the code on the following OS:

• Windows
• MacOS
• Linux

AI Language Model (if applicable)

• GPT4
• GPT3
• Llama 7B
• Llama 13B
• Llama 34B
• Huggingface model (Please specify which one)

[Notnaton]

This might fix this: v0.1.15...v0.1.16#diff-759d9f02bb48ecd82e5b24e5725270c6932e76159967c46ee8742470424dce52

This needs more testing with openai api and others
Need to check code for things that have changed:
https://docs.litellm.ai/docs/migration

What changed?
Requires openai>=1.0.0
openai.InvalidRequestError → openai.BadRequestError
openai.ServiceUnavailableError → openai.APIStatusError
NEW litellm client, allow users to pass api_key
litellm.Litellm(api_key="sk-123")
response objects now inherit from BaseModel (prev. OpenAIObject)
NEW default exception - APIConnectionError (prev. APIError)
litellm.get_max_tokens() now returns an int not a dict

max_tokens = litellm.get_max_tokens("gpt-3.5-turbo") # returns an int not a dict 
assert max_tokens==4097

Streaming - OpenAI Chunks now return None for empty stream chunks. This is how to process stream chunks with content

response = litellm.completion(model="gpt-3.5-turbo", messages=messages, stream=True)
for part in response:
    print(part.choices[0].delta.content or "")

[CyanideByte]

This still causes issues with gpt-4 for me.

(oi) C:\oi>interpreter

▌ Model set to gpt-4-1106-preview

Open Interpreter will require approval before running code.

Use interpreter -y to bypass this.

Press CTRL-C to exit.

hi

    Python Version: 3.11.5
    Pip Version: 23.3
    Open-interpreter Version: cmd:Interpreter, pkg: 0.1.16
    OS Version and Architecture: Windows-10-10.0.19045-SP0
    CPU Info: Intel64 Family 6 Model 94 Stepping 3, GenuineIntel
    RAM Info: 31.95 GB, used: 9.88, free: 22.07


    Interpreter Info
    Vision: False
    Model: gpt-4-1106-preview
    Function calling: True
    Context window: 128000
    Max tokens: 4096

    Auto run: False
    API base: None
    Local: False

    Curl output: Not local

Traceback (most recent call last):
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\pydantic\main.py", line 753, in getattr
return pydantic_extra[item]
~~~~~~~~~~~~~~^^^^^^
KeyError: 'items'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "", line 198, in _run_module_as_main
File "", line 88, in run_code
File "C:\Users\Cyanide\miniconda3\envs\oi\Scripts\interpreter.exe_main.py", line 7, in
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\core.py", line 21, in start_terminal_interface
start_terminal_interface(self)
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\terminal_interface\start_terminal_interface.py", line 305, in start_terminal_interface
interpreter.chat()
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\core.py", line 77, in chat
for _ in self._streaming_chat(message=message, display=display):
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\core.py", line 92, in _streaming_chat
yield from terminal_interface(self, message)
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\terminal_interface\terminal_interface.py", line 115, in terminal_interface
for chunk in interpreter.chat(message, display=False, stream=True):
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\core.py", line 113, in _streaming_chat
yield from self._respond()
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\core.py", line 148, in _respond
yield from respond(self)
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\respond.py", line 49, in respond
for chunk in interpreter._llm(messages_for_llm):
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\llm\setup_openai_coding_llm.py", line 135, in coding_llm
accumulated_deltas = merge_deltas(accumulated_deltas, delta)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\interpreter\core\utils\merge_deltas.py", line 11, in merge_deltas
for key, value in delta.items():
^^^^^^^^^^^
File "C:\Users\Cyanide\miniconda3\envs\oi\Lib\site-packages\pydantic\main.py", line 755, in getattr
raise AttributeError(f'{type(self).name!r} object has no attribute {item!r}') from exc
AttributeError: 'Delta' object has no attribute 'items'

(oi) C:\oi>

[Notnaton]

@CyanideByte thanks, I will check it out after work 👍

[catmeowjiao]

@Notnaton You need to modify all the code that uses the OpenAI module, and I will submit a pull request to solve this problem

[Notnaton]

Excellent @catmeowjiao

[ishaan-jaff]

hi @Notnaton let me know if you run into any issues

[catmeowjiao]

I'm sorry, I've been very busy lately, so the pull request may be delayed. Please forgive me!

[Notnaton]

I'll close this one and make a new, because of out of date. Having some issues with updating too, many changes to the structure.

[vijaykramesh]

any update on this @Notnaton ? I was looking into embedding open-interpreter as a griptape tool but the version conflicts make it difficult..

[Notnaton]

I'm going to look into it again, but it might be after xmas/new years. A lot of people are running into problems with outdated openai/litellm packages.
Maybe tomorrow, no promises @vijaykramesh

[vijaykramesh]

haha I thought I was going insane, I had just set a remote to your fork, was trying to checkout your branch, and suddenly it was gone. Then I refreshed the UI and it also was 404! Fast response 🥇

No worries, look forward to checking back