[BUG] WatsonX LLM BadRequestError: LLM Provider NOT provided. #1583

bishwajitprasadgond · 2024-11-12T11:40:39Z

Description

The error encountered indicates that the LLM provider has not been correctly specified. Specifically, the system is attempting to use the WatsonxLLM model but is not passing the corresponding provider information.

Error Breakdown:
Error Message: BadRequestError: LLM Provider NOT provided.
Details: The model WatsonxLLM was passed, but no provider was specified. The system expects an LLM provider to be provided, such as 'Huggingface', 'Azure', or other supported models.

Steps to Reproduce

Official documentation https://docs.crewai.com/concepts/llms#ibm-watsonx-ai

Parameter sets for each agent type

historical_params = {"decoding_method": "sample", "max_new_tokens": 2000, "temperature": 0.1} # Structured and focused on facts
current_params = {"decoding_method": "greedy", "max_new_tokens": 1500, "temperature": 0.2} # Low variation, focused on precision
report_params = {"decoding_method": "sample", "max_new_tokens": 3500, "temperature": 0.7} # More creative and expansive for summaries
yahoo_params = {"decoding_method": "sample", "max_new_tokens": 2000, "temperature": 0.0} # No variation, focused on precision

Dedicated models for each agent

historical_data_model = "meta-llama/llama-2-13b-chat"
current_data_model = "ibm/granite-13b-instruct-v2"
report_generation_model = "meta-llama/llama-2-13b-chat"

Dedicated models for each agent

historical_data_model = "meta-llama/llama-2-13b-chat"
current_data_model = "meta-llama/llama-2-13b-chat"
report_generation_model = "meta-llama/llama-2-13b-chat"
yahoo_finance_model = "meta-llama/llama-2-13b-chat"

Date range (last two years)

end_date = datetime.now().date()
start_date = datetime(datetime.now().year - 2, 1, 1).date()

Initialize Watsonx LLM instances

historical_data_llm = WatsonxLLM(
model_id=historical_data_model,
url="https://eu-de.ml.cloud.ibm.com",
params=historical_params,
project_id=os.getenv("WATSONX_PROJECT_ID"),
apikey=os.getenv("WATSONX_APIKEY")
)

current_data_llm = WatsonxLLM(
model_id=current_data_model,
url="https://eu-de.ml.cloud.ibm.com",
params=current_params,
project_id=os.getenv("WATSONX_PROJECT_ID"),
apikey=os.getenv("WATSONX_APIKEY")
)

report_writer_llm = WatsonxLLM(
model_id=report_generation_model,
url="https://eu-de.ml.cloud.ibm.com",
params=report_params,
project_id=os.getenv("WATSONX_PROJECT_ID"),
apikey=os.getenv("WATSONX_APIKEY")
)

yahoo_finance_llm = WatsonxLLM(
model_id=yahoo_finance_model,
url="https://eu-de.ml.cloud.ibm.com",
params=yahoo_params,
project_id=os.getenv("WATSONX_PROJECT_ID"),
apikey=os.getenv("WATSONX_APIKEY")
)

Expected behavior

Communicate with LLM

Screenshots/Code snippets

2024-11-12 16:59:14,065 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,075 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,084 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,094 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,102 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,111 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,120 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,131 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,138 - 27388 - llm.py-llm:161 - ERROR: LiteLLM call failed: litellm.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=WatsonxLLM
Params: {'model_id': 'ibm/granite-13b-instruct-v2', 'deployment_id': None, 'params': {'decoding_method': 'sample', 'max_new_tokens': 2000, 'temperature': 0.0}, 'project_id': '4ea20d19-516f-4fdf-9af6-7bd946e6d9b5', 'space_id': None}
Pass model as E.g. For 'Huggingface' inference endpoints pass in completion(model='huggingface/starcoder',..) Learn more: https://docs.litellm.ai/docs/providers
2024-11-12 16:59:14,148 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable

Provider List: https://docs.litellm.ai/docs/providers

Agent: Yahoo Finance Data Specialist

Task: Fetch Yahoo Finance data for IBM, including stock price, P/E ratio, ROE, and recent news. for the timeframe from 2022-01-01 to 2024-11-12.

Provider List: https://docs.litellm.ai/docs/providers

Agent: Yahoo Finance Data Specialist

Task: Fetch Yahoo Finance data for IBM, including stock price, P/E ratio, ROE, and recent news. for the timeframe from 2022-01-01 to 2024-11-12.

2024-11-12 16:59:14,168 - 27388 - llm.py-llm:161 - ERROR: LiteLLM call failed: litellm.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=WatsonxLLM
Params: {'model_id': 'ibm/granite-13b-instruct-v2', 'deployment_id': None, 'params': {'decoding_method': 'sample', 'max_new_tokens': 2000, 'temperature': 0.0}, 'project_id': '4ea20d19-516f-4fdf-9af6-7bd946e6d9b5', 'space_id': None}
Pass model as E.g. For 'Huggingface' inference endpoints pass in completion(model='huggingface/starcoder',..) Learn more: https://docs.litellm.ai/docs/providers
2024-11-12 16:59:14,178 - 27388 - llm.py-llm:178 - ERROR: Failed to get supported params: argument of type 'NoneType' is not iterable
2024-11-12 16:59:14,206 - 27388 - llm.py-llm:161 - ERROR: LiteLLM call failed: litellm.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=WatsonxLLM
Params: {'model_id': 'ibm/granite-13b-instruct-v2', 'deployment_id': None, 'params': {'decoding_method': 'sample', 'max_new_tokens': 2000, 'temperature': 0.0}, 'project_id': '4ea20d19-516f-4fdf-9af6-7bd946e6d9b5', 'space_id': None}
Pass model as E.g. For 'Huggingface' inference endpoints pass in completion(model='huggingface/starcoder',..) Learn more: https://docs.litellm.ai/docs/providers

Provider List: https://docs.litellm.ai/docs/providers

Agent: Yahoo Finance Data Specialist

Task: Fetch Yahoo Finance data for IBM, including stock price, P/E ratio, ROE, and recent news. for the timeframe from 2022-01-01 to 2024-11-12.

BadRequestError Traceback (most recent call last)
File ~\anaconda3\Lib\site-packages\crewai\agent.py:236, in Agent.execute_task(self, task, context, tools)
235 try:
--> 236 result = self.agent_executor.invoke(
237 {
238 "input": task_prompt,
239 "tool_names": self.agent_executor.tools_names,
240 "tools": self.agent_executor.tools_description,
241 "ask_for_human_input": task.human_input,
242 }
243 )["output"]
244 except Exception as e:

File ~\anaconda3\Lib\site-packages\crewai\agents\crew_agent_executor.py:93, in CrewAgentExecutor.invoke(self, inputs)
92 self.ask_for_human_input = bool(inputs.get("ask_for_human_input", False))
---> 93 formatted_answer = self._invoke_loop()
95 if self.ask_for_human_input:

File ~\anaconda3\Lib\site-packages\crewai\agents\crew_agent_executor.py:175, in CrewAgentExecutor._invoke_loop(self, formatted_answer)
174 else:
--> 175 raise e
177 self._show_logs(formatted_answer)

File ~\anaconda3\Lib\site-packages\crewai\agents\crew_agent_executor.py:115, in CrewAgentExecutor._invoke_loop(self, formatted_answer)
114 if not self.request_within_rpm_limit or self.request_within_rpm_limit():
--> 115 answer = self.llm.call(
116 self.messages,
117 callbacks=self.callbacks,
118 )
120 if not self.use_stop_words:

File ~\anaconda3\Lib\site-packages\crewai\llm.py:155, in LLM.call(self, messages, callbacks)
153 params = {k: v for k, v in params.items() if v is not None}
--> 155 response = litellm.completion(**params)
156 return response["choices"][0]["message"]["content"]

File ~\anaconda3\Lib\site-packages\litellm\utils.py:1013, in client..wrapper(*args, **kwargs)
1010 logging_obj.failure_handler(
1011 e, traceback_exception, start_time, end_time
1012 ) # DO NOT MAKE THREADED - router retry fallback relies on this!
-> 1013 raise e

File ~\anaconda3\Lib\site-packages\litellm\utils.py:903, in client..wrapper(*args, **kwargs)
902 # MODEL CALL
--> 903 result = original_function(*args, **kwargs)
904 end_time = datetime.datetime.now()

File ~\anaconda3\Lib\site-packages\litellm\main.py:2999, in completion(model, messages, timeout, temperature, top_p, n, stream, stream_options, stop, max_completion_tokens, max_tokens, modalities, audio, presence_penalty, frequency_penalty, logit_bias, user, response_format, seed, tools, tool_choice, logprobs, top_logprobs, parallel_tool_calls, deployment_id, extra_headers, functions, function_call, base_url, api_version, api_key, model_list, **kwargs)
2997 except Exception as e:
2998 ## Map to OpenAI Exception
-> 2999 raise exception_type(
3000 model=model,
3001 custom_llm_provider=custom_llm_provider,
3002 original_exception=e,
3003 completion_kwargs=args,
3004 extra_kwargs=kwargs,
3005 )

File ~\anaconda3\Lib\site-packages\litellm\main.py:906, in completion(model, messages, timeout, temperature, top_p, n, stream, stream_options, stop, max_completion_tokens, max_tokens, modalities, audio, presence_penalty, frequency_penalty, logit_bias, user, response_format, seed, tools, tool_choice, logprobs, top_logprobs, parallel_tool_calls, deployment_id, extra_headers, functions, function_call, base_url, api_version, api_key, model_list, **kwargs)
905 custom_llm_provider = "azure"
--> 906 model, custom_llm_provider, dynamic_api_key, api_base = get_llm_provider(
907 model=model,
908 custom_llm_provider=custom_llm_provider,
909 api_base=api_base,
910 api_key=api_key,
911 )
912 if model_response is not None and hasattr(model_response, "_hidden_params"):

File ~\anaconda3\Lib\site-packages\litellm\litellm_core_utils\get_llm_provider_logic.py:313, in get_llm_provider(model, custom_llm_provider, api_base, api_key, litellm_params)
312 if isinstance(e, litellm.exceptions.BadRequestError):
--> 313 raise e
314 else:

File ~\anaconda3\Lib\site-packages\litellm\litellm_core_utils\get_llm_provider_logic.py:290, in get_llm_provider(model, custom_llm_provider, api_base, api_key, litellm_params)
289 # maps to openai.NotFoundError, this is raised when openai does not recognize the llm
--> 290 raise litellm.exceptions.BadRequestError( # type: ignore
291 message=error_str,
292 model=model,
293 response=httpx.Response(
294 status_code=400,
295 content=error_str,
296 request=httpx.Request(method="completion", url="https://github.com/BerriAI/litellm"), # type: ignore
297 ),
298 llm_provider="",
299 )
300 if api_base is not None and not isinstance(api_base, str):

BadRequestError: litellm.BadRequestError: LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=WatsonxLLM
Params: {'model_id': 'ibm/granite-13b-instruct-v2', 'deployment_id': None, 'params': {'decoding_method': 'sample', 'max_new_tokens': 2000, 'temperature': 0.0}, 'project_id': '4ea20d19-516f-4fdf-9af6-7bd946e6d9b5', 'space_id': None}
Pass model as E.g. For 'Huggingface' inference endpoints pass in completion(model='huggingface/starcoder',..) Learn more: https://docs.litellm.ai/docs/providers

During handling of the above exception, another exception occurred: