You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by GitBox <gi...@apache.org> on 2020/11/20 06:17:19 UTC

[GitHub] [incubator-tvm] zhxfl opened a new issue #6942: relay.build not support size_var?

zhxfl opened a new issue #6942:
URL: https://github.com/apache/incubator-tvm/issues/6942


   here is the test code 
   ** 80 def test_dropout():
    81     for dtype in ["float16", "float32"]:
    82         #n, t, d = 1024, 1024, 1024
    83         n, t, d = te.size_var("n"), 1024, 1024
    84         input_ty = relay.TensorType((n, t, d), dtype)
    85         x = relay.var("x", input_ty)
    86         y = relay.nn.dropout(x, rate=0.75)
    87         mod = tvm.IRModule.from_expr(relay.Function([x], y))
    88         compiled_lib = relay.build(mod, tvm.target.create("cuda"))
    89         assert "rate=" in y.astext()
    90         yy = run_infer_type(y)
    91         assert yy.checked_type == input_ty
   **
   the graph_plan_memory.cc:GetMemorySize check(dim) fail


----------------------------------------------------------------
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

[GitHub] [incubator-tvm] zhiics closed issue #6942: relay.build not support size_var?

Posted by GitBox <gi...@apache.org>.
zhiics closed issue #6942:
URL: https://github.com/apache/incubator-tvm/issues/6942


   


----------------------------------------------------------------
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

[GitHub] [incubator-tvm] zhxfl commented on issue #6942: relay.build not support size_var?

Posted by GitBox <gi...@apache.org>.
zhxfl commented on issue #6942:
URL: https://github.com/apache/incubator-tvm/issues/6942#issuecomment-732670267


   "Realy VM for dynamic input" can export a lib for "C++" runtime just as "relay.build" ?


----------------------------------------------------------------
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

[GitHub] [incubator-tvm] zhiics commented on issue #6942: relay.build not support size_var?

Posted by GitBox <gi...@apache.org>.
zhiics commented on issue #6942:
URL: https://github.com/apache/incubator-tvm/issues/6942#issuecomment-731281734


   Yes, you have to use Realy VM for dynamic input.


----------------------------------------------------------------
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