add more type support#421
Conversation
yuzawa-san
commented
Jun 8, 2026
yuzawa-san
commented
- add new types, add support for existing types
- try to provide relevant buffer views based on type sizes
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #421 +/- ##
============================================
+ Coverage 94.10% 94.11% +0.01%
Complexity 503 503
============================================
Files 38 38
Lines 1611 1615 +4
Branches 121 121
============================================
+ Hits 1516 1520 +4
Misses 71 71
Partials 24 24
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
nk2IsHere
left a comment
nk2IsHere
left a comment
There was a problem hiding this comment.
It's great! Thank you!
As per float16 being treated as float array in onnx java implementation instead: https://github.com/microsoft/onnxruntime/blob/e8b9b24a31ed791d8cdfc515b2b5dfe46db9d6a6/java/src/main/java/ai/onnxruntime/TensorInfo.java#L341-L344
From our experience we also used float array to represent float16 tensors. If it is to be used with short array as a backing, its would be good to run a small identity model on them to confirm this is what onnx allows to do.
2 participants
