feat(docs): add half-precision training section in using_simulator docs #678
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
PabloCarmona
added
documentation
jubueche
previously approved these changes
Aug 12, 2024
Signed-off-by: Pablo Carmona Gonzalez <[email protected]>
Signed-off-by: Pablo Carmona Gonzalez <[email protected]>
PabloCarmona
force-pushed
the
feat-677-half-precision-docs
branch
from
e017345 to
0877c23
Compare
|
@coreylammie @anu-pub Take a look at it and let me know what you think, thanks! |
|
Compilation is currently not working for the half precision type. To reproduce: Error: |
|
@maljoras I think this is something that you could maybe easily fix. Or you could tell me how I could resolve it. |
maljoras
reviewed
Aug 25, 2024
| dataset = datasets.MNIST("data", train=True, download=True, transform=transform) | ||
| train_loader = torch.utils.data.DataLoader(dataset, batch_size=32) | ||
|
|
||
| model = model.to(device=device, dtype=torch.bfloat16) |
There was a problem hiding this comment.
Does this also work with other tiles that TorchInference? I had added a HALF type, like in the above: rpu_config.runtime.data_type = RPUDataType.HALF
because the HALF could either be bfloat16 of float16 depending on the compilation options.
There was a problem hiding this comment.
There also a runtime.data_type.as_torch() or something function to convert it to the corresponding torch type, see https://github.com/IBM/aihwkit/blob/master/src/aihwkit/simulator/parameters/enums.py#L23
|
Yes, one has to enable the FP16 auto-conversion during the compilation if I remember correctly. You have to make sure that |
|
You could try to set additionally |
|
You might also want to use bfloat as default FP16 type (with |
|
Actually, could also be related to the rather old compiler you are using? One way to go about it might to change each conversion |
|
I suspect the compiler version to be the problem. Do you know what compiler version worked? On the system I using to check this it is not working. Regarding the other things you mentioned: I tried |
|
@jubueche have you tried a different compiler version? what is the status on this testing? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Related issues
#623
#677
Description