Hello, I have encountered a problem when running a glm with the inverse gaussian as distribution and the log as a link function. Do you think this could be a bug, or something else?Attached is a document to run the example. I have written my own functions and they seem to work fine.