[GitHub] [incubator-mxnet] matteosal edited a comment on issue #20149: MXAutogradBackward complains about autograd recording despite being set? (C API)

[GitBox <gi...@apache.org>]

matteosal edited a comment on issue #20149:
URL: https://github.com/apache/incubator-mxnet/issues/20149#issuecomment-817939702


   No luck, by using these flags I still see the same error:
   ```
   const char *cachedop_keys[3] = {"static_alloc", "data_indices", "param_indices"};
   const char *cachedop_vals[3] = {"true", "[0]", "[]"};
   ```
   Besides, I'm trying to mimic the pipeline of `executor.py` and there are no mentions to either `data_indices` or `param_indices` there. 
   By inspecting the source, something related to `AGInfo` entries is created by `MXAutogradMarkVariables`, but only for argument and gradient arrays (which is present in my example). In the `executor.py` interface, such step is performed via the ndarraymethod `attach_grad`, but I can't find an equivalent step for output arrays there.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@mxnet.apache.org
For additional commands, e-mail: issues-help@mxnet.apache.org