automl issues

efficientnetv2-bn parameters for progressive learning

I would like to train one of the efficientnetv2-bn networks on a custom dataset, and I noticed the following in effnetv2_configs.py line 202: > # For fair comparison to EfficientNetV1,...

bahejl

INVALID_ARGUMENT: Input to reshape is a tensor with X values, but the requested shape requires a multiple of Y

2

I'm training `EfficientDet3` on a custom dataset with COCO annotations having `585 categories` and 125 images (as a test, although there are more images), as follows: ``` rm -rf tfrecord...

etale-cohomology

p.add_(..., inplace=True) error

1

commit: c7392f2bab3165244d1c565b66409fa11fa82367 line: automl/lion/lion_pytorch.py:81 When calling `Lion.step()`, I get this error message: > TypeError: add_() received an invalid combination of arguments - got unrecognized keyword arguments: inplace I think `p.add_()`...

markussteindl

buffer_size must be greater than zero error when use custom dataset

1

When I use my custom dataset to train model from scratch, it output the error: > WARNING:tensorflow:`tf.keras.callbacks.experimental.BackupAndRestore` endpoint is deprecated and will be removed in a future release. Please use...

darouwan

how to train model by lion optimizer with fp16?

1

It says "Additional, Lion still requires momentum tracking in bfloat16, which can be expensive for training giant models. One potential solution is to factorize the momentum to save memory.", how...

Geek-lixiang

Question about Lion

1

Thank your for your paper and work, but I meet some wired problem in training. Can you help me? ![image](https://github.com/google/automl/assets/19380638/1ed88f72-0f24-49ef-a6f1-748cf6f7645a) It's my training process curve. Notice Lion will increase loss...

kellenf

f-strings istead of .format method

1

`f-strings` added instead of `.format` method of torch version.

pdumin

Error during prediction within coreML framework of the converted Efficientdet-lite0 model

As described in issue posted on coreML forum: https://developer.apple.com/forums/thread/730922 I successfully finetuned efficientdet-lite0 model, run it correctly on python, converted using coremltools (tensorflow==2.12.0 and coremltools==6.3.0) to mlmodel format and loaded...

Jsbie