Odd issue using Gemini model - chat enters loop

[iainwallacebms]

Very exciting package, thanks

I was able to install it (via remotes rather than pak) and run it with our internal Azure endpoint

        querychat_config <- querychat_init(mtcars,
                                           create_chat_func = purrr::partial(ellmer::chat_azure, 
                                                                       endpoint = Sys.getenv('endpoint'),
                                                                      api_key = Sys.getenv('key'),
                                                                      api_version = "2024-08-01-preview",
                                                                     deployment_id = "gpt-4o-global")

However when I used a gemini model, the chat just goes in a loop and eventually hits a rate limit erroring out.

         querychat_config <- querychat_init(mtcars,
                                       create_chat_func = purrr::partial(ellmer::chat_gemini, 
                                                                      api_key =Sys.getenv('GEMINI_KEY'),
                                                                     model = 'gemini-2.0-pro-exp-02-05'))

[chendaniely]

Hello. Just realized I referenced the wrong issue when I initially tagged #23.

I think the hanging issue was fixed when we updated querychat to match how shinychat handles namespaces in #27 and #10 can you confirm that things are locking up after you install the latest dev versions?