Closed andywenk closed 1 year ago
Hi Andy, do you have access to gpt-4 yet? You can just remove that prompt driver line and it'll default to 3.5 automatically.
yes I have access to gpt-4 but obviously not via API ^^
curl https://api.openai.com/v1/models \
-H "Authorization: Bearer $OPENAI_API_KEY" | grep "gpt-"
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 "id": "gpt-3.5-turbo",
"root": "gpt-3.5-turbo",
"id": "gpt-3.5-turbo-0301",
"root": "gpt-3.5-turbo-0301",
100 43017 100 43017 0 0 54669 0 --:--:-- --:--:-- --:--:-- 54938
May I ask what I need to do to also get access to gpt-4? I created a new key already
That is a good question :) It's unclear how OpenAI is granting access. gpt-3.5-turbo works pretty well for most cases w/ tools attached. If you remove prompt driver, does your code run now against 3.5?
ok thanks - I will try to figure out how to get gpt-4
Erm, basically it works now, but I get another error:
[05/09/23 22:06:56] ERROR Task 37eca2156b9144489f6210d34e86c8fa
0 is less than the minimum of 1 - 'max_tokens'
Traceback (most recent call last):
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/griptape/tasks/base_task.py", line 105, in execute
self.output = self.run()
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/griptape/tasks/toolkit_task.py", line 74, in run
self.active_driver().run(value=self.structure.to_prompt_string(self)).value
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/griptape/drivers/prompt/base_prompt_driver.py", line 32, in run
raise e
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/griptape/drivers/prompt/base_prompt_driver.py", line 25, in run
return self.try_run(**kwargs)
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/griptape/drivers/prompt/openai_prompt_driver.py", line 33, in
try_run
return self.__run_chat(value)
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/griptape/drivers/prompt/openai_prompt_driver.py", line 38, in
__run_chat
result = openai.ChatCompletion.create(
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/openai/api_resources/chat_completion.py", line 25, in create
return super().create(*args, **kwargs)
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/openai/api_resources/abstract/engine_api_resource.py", line 153,
in create
response, _, api_key = requestor.request(
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/openai/api_requestor.py", line 230, in request
resp, got_stream = self._interpret_response(result, stream)
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/openai/api_requestor.py", line 624, in _interpret_response
self._interpret_response_line(
File "/Users/andreaswenk/miniconda/lib/python3.10/site-packages/openai/api_requestor.py", line 687, in _interpret_response_line
raise self.handle_error_response(
openai.error.InvalidRequestError: 0 is less than the minimum of 1 - 'max_tokens'
Okay, that doc wasn't using a Ramp. We have an abstraction called Ramps that helps minimize what is being sent back over the prompt. I updated the example. We are working on getting the docs up to date. We have a new version going out today and we'll update the docs right after that. But, added a ramp to the Agent example you're using.
Just for more context... Ramps are tool wrappers that control data flows between LLMs and the underlying customer data and APIs. Ramps allow for data compression, sanitization, side storage, and obfuscation. Griptape enables LLMs to access ramps directly depending on the context as opposed to relying on the human programmer to connect storage/compression/obfuscation direct (as they have to in other frameworks)
Ok thanks. What do I have to install to make the MemoryStorageDriver available
NameError: name 'MemoryStorageDriver' is not defined
I could not find a pip package
UPDATE: I think this will be in the next release of griptape-tools ...
Arg. from griptape.drivers import MemoryStorageDriver
. I'll make sure that's in the docs.
YAY - that's it ... works great ;-)
Hey again,
thanx for fixing the docs. It works very well. I stumbled into another issue when running the code for Agent:
I would love to check this out by myself and find the error but unfortunately, I am new to this kinda stuff ;-)
Thanks for any hint.
Best
Andy