Hi to all,
I am tracking inputs, outputs and parameters through the Renku
api. I have had this error many times, causing the workflow to not be tracked
finished.
Error: Invalid parameter value - Duplicate input, output or parameter names found: use_embeddings
This is the definition of that parameter:
_ = Parameter(name="use_embeddings", value=args.use_embeddings)
where args.use_embeddings
is a str
. I have tried a few things: cleaning with renku gc
, commenting this (and then it gives the same error with another parameter). But I cannot understand what is going on. In the beginning, as value I was providing True
, and I thought that might be the reason. But now, I just provide a str
. What is going on?
Thanks!
Cheers
Wow, writing this has been a revelation. I feel I know the error. The command is:
renku run python src/bert_topic_extraction.py --nr_topics 300 --use_embeddings False --remove_entities False
Hence, I believe these 3 parameters are tracked twice: internally, and from the command use to run. Hence, I understand the error, but there is still en extremely annoying this: could this be checked before actually running the whole thing? And then renku
prompting: Duplicated inputs, please do something
. Each simulation takes around 1h, and it has been already a few. What are your thoughts on this?
Cheers, and thanks for the inspiration
Thank you for sharing this issue. It will definitely help to improve the error information display.
Have you managed to run the command correctly? (I am sorry that I do not completely get it from your description if you could solve the issue).
HI @elisabetc! Thanks for your response! Yes, I modified the script to not track internally the parameters provided when running the command, i.e. nt_topics
, use_embeddings
and remove_entities
, and then it runs. However, I don’t think something like this should prompt an error, and even less, after the whole execution. For that reason, I do think this behaviour needs to be revised.
Thanks!
Great to hear you solved it . Indeed, I take note to revise this behaviour with the team and avoid it in the future! Thank you again @lusamino