initial working jaxtyping serializing/deserializing #576
Conversation
The code changes in `serde/numpy.py` modify the `is_numpy_jaxtyping` function to handle a `TypeError` exception. This change ensures that if a `TypeError` occurs when trying to determine if a type is a numpy jaxtyping, the function will return `False` instead of raising an error.
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #576 +/- ##
==========================================
- Coverage 91.04% 90.86% -0.18%
==========================================
Files 13 13
Lines 1942 1971 +29
Branches 437 440 +3
==========================================
+ Hits 1768 1791 +23
- Misses 117 122 +5
- Partials 57 58 +1 ☔ View full report in Codecov by Sentry. |
|
Hi @yukinarit, I've implemented initial support for jaxtyping serialization/deserialization and all tests are passing. Before I proceed with adding specific tests for jaxtyping, could you please review the current implementation and let me know if you're happy with the direction? Any feedback would be greatly appreciated! |
|
Made jaxtyping an optional dep and addressed your typ comment! |
|
Added tests in! Let me know if there's anything else you'd like me to add/fix @yukinarit @kmsquire, otherwise I also think this is probably good to merge |
|
@pablovela5620 Is there anything you want to add before publishing v0.20.0? |
|
Nope, this looks good! Thanks |
Jaxtyping serialization/deserializing working along with type checking with beartype #572 . This allows one to use jaxtyping when serializing/deserializing and check both the dtype and the shape. Let me know if this direction makes sense!