You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository was archived by the owner on Jul 10, 2025. It is now read-only.
Copy file name to clipboardExpand all lines: rfcs/20201201-cpp-gradients.md
+58Lines changed: 58 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -251,6 +251,64 @@ Status ExpGradModel(AbstractContext* ctx,
251
251
return Status::OK();
252
252
}
253
253
```
254
+
**TapeContext**
255
+
256
+
In the final C++ API, we don’t expect users to have to directly call `RecordOperation` for each op. We would provide an `AbstractContext` implementation for the tape which would trace ops on the tape and delegate execution to a backing context. A prototype for this is available [here](https://cs.opensource.google/tensorflow/tensorflow/+/master:tensorflow/c/experimental/gradients/tape/).
0 commit comments