Hi @Curtis_Holloway!
I wanted to check in and see if you still needed help resolving this?
Happy to help in any way!
Hello Ford, yes I still need help with this issue, It is persistent. If I did something wrong, I don’t know what it would be.
Hi @Curtis_Holloway!
I’m sorry that you have run into this! Do you mind providing further details and screenshots to help me troubleshoot?
It just gives me the generic issue that you see in the screenshot below. I don’t have any other information.
Hi @Curtis_Holloway!
Do you give me permission to access your workspace? I can take a look and greatly expedite the triage process. Thank you!!
Yes please. I need to get this project working
Hi @Curtis_Holloway!
Thank you for your patience, I wanted to provide you an update. I was able to recreate the limitations you ran into and escalated it to the appropriate team to be addressed.
I will keep you updated on our progress. Thank you again for bringing this to our attention!!
Thanks Ford, I look forward to hearing from you soon so I can continue my project!
Hi @Curtis_Holloway!
Thank you for your continued patience. Our team is still investigating your issue.
I will keep you updated.
Hi Curtis,
I took a look at your workspace / project and found the error logs from when you try to generate versions.
Your dataset has 27000+ classes, this is hitting in indexing limit on our database that I’m not sure we can work around. Even if we could I dont think you’re going to be able to train a model that produces meanigful results with that many classes.
It looks like you are trying to identify specific objects out of a large set of inventory. I think you might be better served with an embedding approach; i.e. you could use an embedding model like CLIP or Perception Encoder to compute embedding vectors for all your examples, store them in a vector db like e.g. qdrant (qdrant.tech) and then when you need have an image you want to look up compute the embeddings for it and do a search against the vector DB.
Ok that is what I needed to know. Thanks