earth-forecasting-transformer

earth-forecasting-transformer copied to clipboard

how to Hyperparameters of the AdamW optimizer? How to use global vectors?

4

when I train the ICAR-ENSO dataset,I found the hyperparameter settings of AdamW in the paper are different from the parameter settings in the earthformer_enso_v1.yaml file, so which one should be...

Yang-Changhui

Compatibility issues for pytorch_lightning

1

Taking the N-body minist script as an example, the code can't be executed with pytorch 2.0.1 due to the compatibility issue of pytorch_lightning. Is there any workarounds to bypass these...

qiyang77

Question about the 2D CNN+Downsampler

1

Thanks for your excellent contribution! I'm trying to understand the structure of CuboidTransformerModel, but I find the implementation is not consistent with the description in the paper. According to **Table...

ycshao21

Both use the sevir dataset to predict for two hours Using the adamw optimizer with a learning rate of 1e-4, none of them used the task learning rate optimization strategy...

Helomin

Memory issues with the earthformer runtime

1

I set the dimension as (2000,131,21,1), that is, (T,H,W,C) the data input model can run, (the monitor shows that the runtime code takes up 13GB), but when increasing the timing,...

fizzking

Questions about the target_shape of the enso dataset

5

The label of enso is Nino3.4 SST anomaly index, and the data dimension is (year,month). But why targe shape in cfg is a four-dimensional array of (12,24, 48,1), shouldn't it...

fizzking

Question about hierarchical Structure

1

Hi, researching your paper I read that the used architecture utilizes a hierarchical Structure in a similar way as in the Hourglass transformer ( https://arxiv.org/abs/2110.13711 ). But for your case...

Titolasanta

How to run my own datasetwithin Earthformer

11

I want to use Earthformer to train my own dataset and test it, what format should I process the data into and what py files should I prepare?

fizzking

RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

5

Traceback (most recent call last): File "/root/autodl-tmp/earth-forecasting-transformer/scripts/cuboid_transformer/sevir/train_cuboid_sevir.py", line 792, in main() File "/root/autodl-tmp/earth-forecasting-transformer/scripts/cuboid_transformer/sevir/train_cuboid_sevir.py", line 781, in main trainer.fit(model=pl_module, File "/root/miniconda3/envs/earthformer/lib/python3.9/site-packages/pytorch_lightning/trainer/trainer.py", line 770, in fit self._call_and_handle_interrupt( File "/root/miniconda3/envs/earthformer/lib/python3.9/site-packages/pytorch_lightning/trainer/trainer.py", line 721, in...

Happiyin

Question about the input shape of 'CuboidTransformerEncoder' in cuboid_transformer.py

1

I am currently working with multiband (4 band) satellite images (B, T, H, W, C=4) and I have a question regarding the code in line 2930: In line 2930: self.encoder...

paradigm21c