Fix max_token=2 for togetherAI

[yujonglee]

tests/evaluation/test_with_yelp_review.py::test_llm_grading_head[togethercomputer/llama-2-70b-chat-references1]
  /Users/yujonglee/dev/fastrepl/fastrepl/fastrepl/warnings.py:24: UnknownLLMExceptionWarning: ValueError: Traceback (most recent call last):
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/utils.py", line 1755, in exception_type
      error_response = json.loads(error_str)
                       ^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/.pyenv/versions/3.11.3/lib/python3.11/json/__init__.py", line 346, in loads
      return _default_decoder.decode(s)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/.pyenv/versions/3.11.3/lib/python3.11/json/decoder.py", line 337, in decode
      obj, end = self.raw_decode(s, idx=_w(s, 0).end())
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/.pyenv/versions/3.11.3/lib/python3.11/json/decoder.py", line 355, in raw_decode
      raise JSONDecodeError("Expecting value", s, err.value) from None
  json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

  During handling of the above exception, another exception occurred:

  Traceback (most recent call last):
    File "/Users/yujonglee/dev/fastrepl/fastrepl/fastrepl/llm.py", line 138, in _completion
      result = litellm.gpt_cache.completion(  # pragma: no cover
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/gptcache/adapter/openai.py", line 100, in create
      return adapt(
             ^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/gptcache/adapter/adapter.py", line 238, in adapt
      llm_data = time_cal(
                 ^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/gptcache/utils/time.py", line 9, in inner
      res = func(*args, **kwargs)
            ^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/gpt_cache.py", line 12, in _llm_handler
      return litellm.completion(*llm_args, **llm_kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/utils.py", line 565, in wrapper
      raise e
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/utils.py", line 526, in wrapper
      result = original_function(*args, **kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/timeout.py", line 44, in wrapper
      result = future.result(timeout=local_timeout_duration)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/.pyenv/versions/3.11.3/lib/python3.11/concurrent/futures/_base.py", line 456, in result
      return self.__get_result()
             ^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/.pyenv/versions/3.11.3/lib/python3.11/concurrent/futures/_base.py", line 401, in __get_result
      raise self._exception
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/timeout.py", line 33, in async_func
      return func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/main.py", line 825, in completion
      raise exception_type(
            ^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/utils.py", line 1826, in exception_type
      raise original_exception
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/main.py", line 559, in completion
      model_response = together_ai.completion(
                       ^^^^^^^^^^^^^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/litellm/llms/together_ai.py", line 110, in completion
      model_response["choices"][0]["message"]["content"] = completion_response
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^
    File "/Users/yujonglee/dev/fastrepl/fastrepl/.venv/lib/python3.11/site-packages/openai/openai_object.py", line 71, in __setitem__
      raise ValueError(
  ValueError: You cannot set content to an empty string. We interpret empty strings as None in requests.You may set {
    "content": "default",
    "role": "assistant",
    "logprobs": null
  }.content = None to delete the property
   | https://docs.fastrepl.com/miscellaneous/warnings_and_errors#unknownllmexception

This is not problem with LiteLLM side.

[ishaan-jaff]

this is because the response from tg ai was empty btw