convert_to_model_input() builds json_tensors list but never includes it in the return value (dead data, silent data loss)

[verseon0980]

File: src/opengradient/client/_conversions.py lines 60, 112

The function initialises json_tensors = [] on line 60 but the return statement on line 112 only returns (number_tensors, string_tensors)

The json_tensors list is never populated no elif branch for JSON dtype and is never returned.

This is entirely dead code,

But it sets up a future trap: if someone adds JSON tensor support to the collection loop without fixing the return statement, JSON inputs will silently disappear.

Current return:
pythonreturn number_tensors, string_tensors

Remove the dead variable. If JSON tensor input is intended:

pythonreturn number_tensors, string_tensors, json_tensors

and update all callers accordingly.

[teyrebaz33]

Hey, I've opened PR #259 to address this.

Went with the clean removal approach — since there's no JSON tensor collection logic or caller support, removing the dead variable is the safest fix. If JSON tensor support is added in the future, it should come with the full implementation in a single PR.

Happy to take a different approach if preferred!

[verseon0980]

Good to see you are doing the PR
I hope they get fixed