ALinrunrun opened a new issue, #19561:
URL: https://github.com/apache/tvm/issues/19561
### Expected behavior
TVM Relax should execute ONNX `Asinh` consistently with ONNX Runtime for
large finite float32 inputs.
For large finite values, `asinh(x)` should remain finite and approximately
grow like `log(2x)`. Negative inputs should produce negative outputs.
### Actual behavior
TVM Relax returns `inf` for all elements, while ONNX Runtime returns finite
values:
```
input: [ 3.e+22 5.e+25 -8.e+23]
ORT : [ 52.44863 59.86721 -55.732044]
TVM : [inf inf inf]
```
The discrepancy appears when importing an ONNX Asinh model through the Relax
ONNX frontend and compiling it for the llvm target.
### Environment
TVM: 0.14 environment / Relax ONNX frontend
ONNX Runtime: 1.23
Python: 3.11
Target: llvm
OS: Linux
### Steps to reproduce
```
import warnings
warnings.filterwarnings("ignore")
import numpy as np
import onnxruntime as ort
import tvm
from onnx import TensorProto, helper
from tvm import relax
from tvm.relax.frontend.onnx import from_onnx
node = helper.make_node("Asinh", ["x"], ["y"])
graph = helper.make_graph(
[node],
"g",
[helper.make_tensor_value_info("x", TensorProto.FLOAT, [3])],
[helper.make_tensor_value_info("y", TensorProto.FLOAT, [3])],
)
model = helper.make_model(graph, opset_imports=[helper.make_opsetid("", 20)])
model.ir_version = 9
x = np.array([3e22, 5e25, -8e23], dtype=np.float32)
ort_out = ort.InferenceSession(
model.SerializeToString(),
providers=["CPUExecutionProvider"],
).run(None, {"x": x})[0]
mod = from_onnx(model)
with tvm.transform.PassContext(opt_level=3):
ex = tvm.compile(mod, target=tvm.target.Target("llvm"))
vm = relax.VirtualMachine(ex, tvm.cpu())
out = vm["main"](tvm.runtime.tensor(x, tvm.cpu()))
tvm_out = (out[0] if isinstance(out, (list, tuple)) else out).numpy()
print("input:", x)
print("ORT :", ort_out)
print("TVM :", tvm_out)
```
### Triage
* needs-triage
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
