Error Instancing bedrock model from example code about scrapegraph-ai HOT 2 CLOSED

The way I fixed this is by removing the temperature parameter. However, somehow, somewhere in the pipeline, the instantiation of the bedrock model turns into a string, and I get an error about 'str' object has no attribute .invoke_model.

I will fix this next and raise an issue about it.

from scrapegraph-ai.

Okay, here is the problem.

In this chunk of code:

graph_config = {
"llm": {
"client": "client_name",
"model": "bedrock/anthropic.claude-3-sonnet-20240229-v1:0"
}
}

"client" needs to map to a boto3 instance of bedrock client. The langchain library for aws has the method init_chat_model(), which will not throw an error if you pass in a string for the client, but it will throw an error when trying to use llm_model.generate_prompt() saying that "AttributeError: 'str' object has no attribute 'to_messages'"

I am going to keep going down this rabbit hole to see if I can solve this and continue reporting on this thread.

Edit: Unfortunately, this is incorrect, and instantiating "client_name" as the actual instance of the boto3 bedrock client did not fix it. I would appreciate documentation on what client name is.

from scrapegraph-ai.