In an ideal world we'd never have to deal with these, but being a technical product, that idea is a mere, mere fantasy. If you do get one or more of these though, all hope is not lost! They can almost always be fixed with a little bit of magic...and/or code debugging.
Some of the more common errors are listed below but you may also want to check out the full list of status codes on our Developer Site.
Failure
Error code: 10020
This is a high level error code and an extremely common one that is basically saying that we are unable to retrieve data that you are trying to send over to us (closely related to error code 30300 below, which explains it in more detail). You should be receiving a secondary code if you dig deeper in the message, but usually a 10020 means that you're sending:
- A non-public URL (from AWS, for example)
- A corrupt file
- A format that we don't support
- A duplicate URL that is already in your collection (which is set to error by default, but is override-able), or
- A file hosted on a server far, far away that makes it really tough for us download it in a timely fashion
clarifai.client.client.ApiError: 'UNAUTHORIZED'
Error code: 11002
This almost always means that there is something wrong with your API Key credentials. Make sure that they're written correctly and attached to your application.
Input image decoding failed. check URLs and bytes sent
Error code: 30300
This means that there's a slight problem with retrieving the data that you sent over to us. The most common causes of this are:
- A non-public URL (from AWS, for example)
- A corrupt file
- A format that we don't support
- A file hosted on a server far, far away that makes it really tough for us download it in a timely fashion
If it's the last one, then hosting on a different server would help if it was closer to the U.S. east coast, as that is where our servers are.
Model training had no positive examples
Error code: 21111
This is our system's way of telling you that one or more concepts in your custom model are essentially "zombies" and not tied to any inputs. Once a concept is added to a model it must be used somewhere - else the system will complain about it.
The good news about this one is that it will tell you exactly which concepts are zombied. At least it should be. If it's not for some reason then please let us know!
Internal Error. Contact Support
Error code: 99009
Ah, that's a fun one isn't it? When our system is unable to determine what a specific error is, it'll give you this generic one that isn't a ton of help, admittedly. Any error codes that start with "99" are indicative of a possible server problem and should be reported to us.
Account limits exceeded: Request exceeds "custom concept" limit
Error code: 11006
There are several things that can cause an account limit to be reached and this is one of the more common ones. Custom Model Concepts are actually counted across all model versions (not just your latest one) and thus deleted ones won't necessarily be subtracted from that, unless you delete the old model version that contained them. This can either be done with our User Interface or our API and the instructions are below!
https://help.clarifai.com/hc/en-us/articles/360038139053-How-to-Delete-Old-Custom-Model-Versions
Resource does not exist. Model_id 'X' not found. Check the url of your request.
Error code: 11101, 11103
This one may seem a bit scary at first since your model seems like it's ghosting from our platform, but it's really just us looking for the model name here. Thus, if you named your model "Crazy Partying Koalas", your request should look something like:
cURL:
https://api.clarifai.com/v2/models/Crazy%20Partying%20Koalas/outputs
Python:
model = app.models.get("Crazy Partying Koalas")
...etc...
No GDPR Consent. User doesn't have GDPR Consent.
Error code: 11104
Simply put, this means that you haven't consented to our GDPR regulations yet, which means that you'll need to log out, log back in and then click on the appropriate checkboxes that appear.
If you don't see these checkboxes for whatever reason please let us know!