When running the CLI reginald_run to set up slack bot with a llama-index model, you get error when you don't have the index built (i.e. the correct folder doesn't exist in your data directory) in which case you need to build the index by using --force-new-index flag. We could have better error handling.
Either:
Build it if you don't have it - so not necessary to --force-new-index (it'd still be useful to have because there's times where it does exist and you need to build again)
Give better error handling to let the user know that the folder doesn't exist and need to --force-new-index
When running the CLI
reginald_run
to set up slack bot with allama-index
model, you get error when you don't have the index built (i.e. the correct folder doesn't exist in your data directory) in which case you need to build the index by using--force-new-index
flag. We could have better error handling.Either:
--force-new-index
(it'd still be useful to have because there's times where it does exist and you need to build again)--force-new-index
The specific lines of code to change are here: https://github.com/alan-turing-institute/reginald/blob/1f54b29a05318b876cef0ba1ea878bbed99da0c1/reginald/models/models/llama_index.py#L146-L168
Note: consider the folder may not exist because the data directory was misspecified, so perhaps the first suggestion is not ideal.