slyubomirsky opened a new issue #7466:
URL: https://github.com/apache/tvm/issues/7466


   The following program does not type check and arguably shouldn't, but the 
error message is an internal check failure rather than a proper error message 
stating what the error is
   
   ```
   import tvm
   from tvm import relay
   mod = tvm.IRModule()
   mod["main"] = relay.Function([], relay.nn.bias_add(relay.const(1), 
relay.const(2), axis=0))
   mod = relay.transform.InferType()(mod)
   ```
   
   The error message is:
   ```
   ---------------------------------------------------------------
   An internal invariant was violated during the execution of TVM.
   Please read TVM's error reporting guidelines.
   More details can be found here: 
https://discuss.tvm.ai/t/error-reporting/7793.
   ---------------------------------------------------------------
     Check failed: false == false: [21:00:55] 
/home/sslyu/tvm/include/tvm/runtime/container.h:683:
   ---------------------------------------------------------------
   An internal invariant was violated during the execution of TVM.
   Please read TVM's error reporting guidelines.
   More details can be found here: 
https://discuss.tvm.ai/t/error-reporting/7793.
   ---------------------------------------------------------------
     Check failed: 0 <= i && i < p->size_ == false: IndexError: indexing 0 on 
an array of size 0
   ```
   
   The cause is indexing into the input shape at [this 
line](https://github.com/apache/tvm/blob/b7e0cfb6d469c3745ae2195908daadea9c64d87e/src/relay/op/nn/nn.cc#L68),
 but there should probably be a more informative error message than this. (I 
was just looking at some type relations when I noticed this. I imagine there 
are probably also failure cases that should be handled more gracefully.)


----------------------------------------------------------------
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:
[email protected]


Reply via email to