Very odd error happened


#1

I checked out the tvm in Jun on Ubuntu, and has not checked out from github of tvm again. these days, I checked out the recent version of tvm for Windows. Today, I tried old python scirpt from Ubuntu, very odd error happened.
net, params = relay.frontend.from_tensorflow(graph_def, layout=layout, shape={‘preprocess/truediv’: input_shape})

tasks = autotvm.task.extract_from_program(net, target=target, params=params, ops=(relay.op.nn.conv2d,)) when runned to this line, an error happened as below:

File “so_check_mmnet_input_384.py”, line 263, in
net, params = tvm.relay.testing.create_workload(net)

File “C:\Users\yl2304\tvm\python\tvm\relay\testing\init.py”, line 153, in create_workload
mod = relay.Module.from_expr(net)

File “C:\Users\yl2304\tvm\python\tvm\relay\module.py”, line 204, in from_expr
return _module.Module_FromExpr(expr, funcs, defs)

File “C:\Users\yl2304\tvm\python\tvm_ffi_ctypes\function.py”, line 210, in call
raise get_last_ffi_error()

tvm._ffi.base.TVMError: [18:00:25] f:\learning\code\tvm\include\tvm\packed_func_ext.h:143: Check failed: NodeTypeChecker::Check(sptr.get()): Expected type relay.Expr but get relay.Module

Is there any big API changed since June?

Thanks a lot!


#2

frontend api now returns relay.Module, the original function can be accessed via module[‘main’]