jdeschamps
commented
3 months ago Hi @drchrisch,
Thanks for the feedback!!
path is the parameter that will control the name of the zip file. In your example, it would yield n2v_models.zip.
Error
For completeness here is the error that you probably encountered:
pydantic_core._pydantic_core.ValidationError: 2 validation errors for bioimage.io model specification
name
Value error, 'n2v_model_example.bmz' is not restricted to 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_- ()' [type=value_error, input_value='n2v_model_example.bmz', input_type=str]
For further information visit https://errors.pydantic.dev/2.7/v/value_error
authors.0.email
value is not a valid email address: The part after the @-sign is not valid. It should have a period. [type=value_error, input_value='nobody@nobody', input_type=str]Reason
Pydantic is the validation library used in both CAREamics and the BMZ core library. What Pydantic tells us here is that there are two parameters that are wrong:
name, because it should only consist of letters, numbers, hyphens, underscores, spaces and parentheses.authors.0.email, because it is missing a period after the @-sign.
Indeed, we see that in your code snipper, name=n2v_model_example.bmz (which has a forbidden sign, namely .) and "email": "nobody@nobody" (which is missing a dot after the @-sign).
Fix
By fixing both these parameters, the code should run:
import numpy as np
from careamics import CAREamist
from careamics.config import create_n2v_configuration
# create a configuration
config = create_n2v_configuration(
experiment_name="n2v_2D",
data_type="array",
axes="YX",
patch_size=[64, 64],
batch_size=1,
num_epochs=1, # (1)!
)
# instantiate a careamist
careamist = CAREamist(config)
# train the model
train_data = np.random.randint(0, 255, (256, 256)).astype(np.float32) # (2)!
careamist.train(train_source=train_data)
# once trained, predict
pred_data = np.random.randint(0, 255, (128, 128)).astype(np.float32)
predction = careamist.predict(source=pred_data)
careamist.export_to_bmz(
path="n2v_models",
name="n2v_model_example_bmz",
input_array=train_data,
authors=[{ "name": "nobody", "email": "nobody@nobody.nb", }]
)Note that the line with the # (2)! has now .astype(np.float32, otherwise you will run into another error.
Documentation
This example was added to the FAQ and a comment has been added on the name parameter in the export to BMZ page.
We will update the method documentation in the library itself.
drchrisch
Describe the bug I tried the simple n2v example "Basic CAREamics usage" from [https://careamics.github.io/0.1/guides/careamist_api/]. This worked just as expected. Then, I wanted to save the generated model using the "export_to_bmz" function and ended up with lots of error messages.
To Reproduce Simply run the code from "Basic CAREamics usage" and add
careamist.export_to_bmz( path="n2v_models", name="n2v_model_example.bmz", input_array=train_data, authors=[{ "name": "nobody", "email": "nobody@nobody", }] )Expected behavior I was expecting to get a zip (or bmz) file containing model for further use.
Environment:
Following instructions from [https://careamics.github.io/0.1/installation/]: mamba create -n careamics python=3.10 mamba activate careamics mamba install pytorch torchvision pytorch-cuda=11.8 -c pytorch -c nvidia