👋 Hello, team! I am working with Wren AI for generating question recommendations using the Hamilton framework and the Ollama generator for LLM responses. While executing the pipeline, I encountered a KeyError: 'response', causing the process to fail. Here are my questions: What might cause a KeyError: 'response' during Wren AI's question generation using Ollama? Are there specific configurations I should verify? How can I debug or log more details in Hamilton when encountering issues with the Ollama generator's response structure? Are there known compatibility issues between Wren AI and Ollama that might lead to task failures or timeouts in generate() pipelines? Regards

Hey! So looking at the logs this appears to be an issue inside wren with ollama — Hamilton is just the orchestration component here.

Thank you for the details! I'll explore the logging options and the Hamilton UI to gain more insights into what's happening during the generate() node execution. If there are specific observability features in Wren/Ollama that I should focus on, let me know when you're available. Appreciate your help and will wait for further guidance!