Fix for issue #213 #289
Fix for issue #213 #289
Conversation
torchquantum/operator/op_types.py
Outdated
| from ..macro import C_DTYPE, F_DTYPE | ||
| from typing import Iterable, Union, List | ||
| from enum import IntEnum | ||
| from torchquantum.util.quantization.clifford_quantization import CliffordQuantizer |
There was a problem hiding this comment.
Is this import necessary here?
There was a problem hiding this comment.
Oh, wait, I missed your initial comment.
There was a problem hiding this comment.
Can you instead import this lazily, so in the conditional branch where we actually call CliffordQuantizer? Afterward, will merge in!
01110011011101010110010001101111
left a comment
01110011011101010110010001101111
left a comment
There was a problem hiding this comment.
Thanks for figuring this out! Just a quick question on the extra import then I can merge this in!
torchquantum/operator/op_types.py
Outdated
| from ..macro import C_DTYPE, F_DTYPE | ||
| from typing import Iterable, Union, List | ||
| from enum import IntEnum | ||
| from torchquantum.util.quantization.clifford_quantization import CliffordQuantizer |
There was a problem hiding this comment.
Oh, wait, I missed your initial comment.
torchquantum/operator/op_types.py
Outdated
| from ..macro import C_DTYPE, F_DTYPE | ||
| from typing import Iterable, Union, List | ||
| from enum import IntEnum | ||
| from torchquantum.util.quantization.clifford_quantization import CliffordQuantizer |
There was a problem hiding this comment.
Can you instead import this lazily, so in the conditional branch where we actually call CliffordQuantizer? Afterward, will merge in!
|
Also, don’t worry about tests, it’s likely from numpy 2.0 breaking. Once we merge, I think we should be good. |
|
Alright I moved the import into the conditional branch, so it should be good now. |
|
Merged, thanks so much! |
2 participants
To fix issue #213, it looks like the
QuantumDeviceneeds to be initialized with the batch size (batch size is 1 by default which is why the size was [1, 2] instead of [256, 2]).After this change, there is a
NameErrorforCliffordQuantizer:which can be fixed by importing it in the
torchquantum/operator/op_types.pyfile, which calls theCliffordQuantizerfunction.After these changes, I believe
examples/clifford_qnn/mnist_clifford_qnn.pycan run out-of-the-box:@Hanrui-Wang @01110011011101010110010001101111