training config #6
Description
Traceback (most recent call last):
File "/mnt/tmp/HiCo_T2I/train_hico.py", line 1311, in
main(args)
File "/mnt/tmp/HiCo_T2I/train_hico.py", line 997, in main
train_dataset = make_train_dataset_coco(cfg_data, 'train', accelerator)
File "/mnt/tmp/HiCo_T2I/train_hico.py", line 735, in make_train_dataset_coco
dataset = build_coco_dsets(cfg_data, mode=mode)
File "/mnt/tmp/HiCo_T2I/dataset/coco.py", line 495, in build_coco_dsets
deprecated_coco_stuff_ids_txt=os.path.join(params.root_dir, params[mode].deprecated_stuff_ids_txt),
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/dictconfig.py", line 375, in getitem
self._format_and_raise(key=key, value=None, cause=e)
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/base.py", line 231, in _format_and_raise
format_and_raise(
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/_utils.py", line 899, in format_and_raise
_raise(ex, cause)
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/_utils.py", line 797, in _raise
raise ex.with_traceback(sys.exc_info()[2]) # set env var OC_CAUSE=1 for full trace
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/dictconfig.py", line 369, in getitem
return self._get_impl(key=key, default_value=DEFAULT_MARKER)
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/dictconfig.py", line 442, in _get_impl
node = self._get_child(
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/basecontainer.py", line 73, in _get_child
child = self._get_node(
File "/home/nsml/miniconda3/envs/hico/lib/python3.10/site-packages/omegaconf/dictconfig.py", line 480, in _get_node
raise ConfigKeyError(f"Missing key {key!s}")
omegaconf.errors.ConfigKeyError: Missing key train
full_key: data.parameters.train
object_type=dict
**I'm getting an error saying that the train key is missing in the dataset configuration file. How should the dataset be prepared?**
Is the following way of preparing the crop location correct?
python
r_obj_class.insert(0, caption)
r_obj_bbox = np.insert(r_obj_bbox, obj=0, values=[0, 0, 512, 512], axis=0)
cond_image = np.zeros_like(r_image, dtype=np.uint8)
list_cond_image = []
cond_image = np.zeros_like(r_image, dtype=np.uint8)
list_cond_image.append(cond_image)
for iit in range(1, len(r_obj_bbox)):
dot_bbox = r_obj_bbox[iit]
dx1, dy1, dx2, dy2 = [int(xx) for xx in dot_bbox]
cond_image = np.zeros_like(r_image, dtype=np.uint8)
cond_image[dy1:dy2, dx1:dx2] = 255
list_cond_image.append(cond_image)
obj_cond_image = np.stack(list_cond_image, axis=0)
Additionally:
How can I confirm that cond_image is a blank image?