'Google' object has no attribute 'co' and 'AttributeError' object has no attribute 'message'

[aparnajore]

After resolving the issue #417, I tried to use Google LM but facing some new issues like
AttributeError: 'Google' object has no attribute 'co'
AttributeError: 'AttributeError' object has no attribute 'message'

---

AttributeError Traceback (most recent call last)
/usr/local/lib/python3.10/dist-packages/backoff/_sync.py in retry(*args, **kwargs)
104 try:
--> 105 ret = target(*args, **kwargs)
106 except exception as e:

8 frames
/usr/local/lib/python3.10/dist-packages/dsp/modules/google.py in request(self, prompt, **kwargs)
94 """Handles retrieval of completions from Google whilst handling API errors"""
---> 95 return self.basic_request(prompt, **kwargs)
96

/usr/local/lib/python3.10/dist-packages/dsp/modules/google.py in basic_request(self, prompt, **kwargs)
73 }
---> 74 response = self.co.generate(**kwargs)
75

AttributeError: 'Google' object has no attribute 'co'

During handling of the above exception, another exception occurred:

AttributeError Traceback (most recent call last)
in <cell line: 18>()
16 # Call the predictor on a particular input.
17 question='What is the color of the sky?'
---> 18 pred = generate_answer(question=question)
19
20 print(f"Question: {question}")

/usr/local/lib/python3.10/dist-packages/dspy/predict/predict.py in call(self, **kwargs)
47
48 def call(self, **kwargs):
---> 49 return self.forward(**kwargs)
50
51 def forward(self, **kwargs):

/usr/local/lib/python3.10/dist-packages/dspy/predict/predict.py in forward(self, **kwargs)
88
89 if self.lm is None:
---> 90 x, C = dsp.generate(template, **config)(x, stage=self.stage)
91 else:
92 # Note: query_only=True means the instructions and examples are not included.

/usr/local/lib/python3.10/dist-packages/dsp/primitives/predict.py in do_generate(example, stage, max_depth, original_example)
76 # Generate and extract the fields.
77 prompt = template(example)
---> 78 completions: list[dict[str, Any]] = generator(prompt, **kwargs)
79 completions: list[Example] = [template.extract(example, p) for p in completions]
80

/usr/local/lib/python3.10/dist-packages/dsp/modules/google.py in call(self, prompt, only_completed, return_sorted, **kwargs)
102 **kwargs,
103 ):
--> 104 return self.request(prompt, **kwargs)

/usr/local/lib/python3.10/dist-packages/backoff/_sync.py in retry(*args, **kwargs)
109 elapsed >= max_time_value)
110
--> 111 if giveup(e) or max_tries_exceeded or max_time_exceeded:
112 _call_handlers(on_giveup, **details, exception=e)
113 if raise_on_giveup:

/usr/local/lib/python3.10/dist-packages/dsp/modules/google.py in giveup_hdlr(details)
23 def giveup_hdlr(details):
24 """wrapper function that decides when to give up on retry"""
---> 25 if "rate limits" in details.message:
26 return False
27 return True

AttributeError: 'AttributeError' object has no attribute 'message'

[aparnajore]

I tried to resolve above issues and generate a response but getting only single character in the output and the completions gets omitted after that. But when tried with GPT3, I am getting complete response. Could you please help me for getting complete response using Google LM?

[arnavsinghvi11]

tagging @CShorten here. I think self.co is supposed to be self.google right? :)

That might fix up the error since I don't believe there's a generation yet.

[markgunnels]

I'm getting this error as well. I don't think genai.configure returns anything. I think self.google should maybe be set to genai.GenerativeModel with the model passed as an argument after genai.configure is called?

I'm a Python newbie. YMMV.

[CShorten]

Hey @markgunnels, nice to see you here!

Thanks for tagging @arnavsinghvi11 and for finding this @aparnajore! This PR will fix it -- please note the google.generativeai changed significantly from last week, and I think it makes sense to change the naming a bit as well with GoogleVertex and GoogleGenerativeAI.

#472

[CShorten]

This is a better version -- #461. Just tested, worked for me!

[aparnajore]

Hello @CShorten, thanks for the updates. Could you please let me know when it will be reflected in main code?

[aparnajore]

I tried with #461 but getting InternalServerError
InternalServerError: 500 POST http://localhost:39955/v1beta/models/gemini-1.0-pro:generateContent?%24alt=json%3Benum-encoding%3Dint: TypeError: Failed to fetch.

[nbqu]

Google API sometimes raises 500 errors, so you need to handle these failures by specifying max_tries in backoff decorator.

Hope you can resolve this problem!

[aparnajore]

Thanks @CShorten and @nbqu, now its working.