DeepReg: error after latest git pull (main branch)

getting this error while launching a training script which was previously working (with the same configuration file):

Traceback (most recent call last):
  File "demos/paired_MMIV/demo_train.py", line 29, in <module>
    log_dir=log_dir,
  File "/home/charlie/3DREG-tests/DeepReg/deepreg/train.py", line 94, in train
    max_epochs=max_epochs,
  File "/home/charlie/3DREG-tests/DeepReg/deepreg/train.py", line 45, in build_config
    config = config_parser.load_configs(config_path)
  File "/home/charlie/3DREG-tests/DeepReg/deepreg/parser.py", line 41, in load_configs
    config_sanity_check(config)
  File "/home/charlie/3DREG-tests/DeepReg/deepreg/parser.py", line 90, in config_sanity_check
    if config["train"]["method"] == "conditional":
KeyError: 'method'

the script is

from deepreg.train import train
import sys

######## TRAINING ########

gpu = "2,3"
gpu_allow_growth = True
log_dir = "paired_logs_train"

config_path = [
    r"demos/paired_MMIV/paired_MMIV_train.yaml",
    r"demos/paired_MMIV/paired_MMIV.yaml",
]

if len(sys.argv) != 2:
    print('using training log dir ', log_dir)
else:
    log_dir = "paired_logs_train_"+sys.argv[1]
    print('using training log dir ', log_dir)

#ckpt_path = "logs/paired_logs_train_10000epochs_gmi_unet_lw1rw1_learnrate1e-5/save/weights-epoch10000.ckpt" # restart from saved model
ckpt_path = ""

train(
    gpu=gpu,
    config_path=config_path,
    gpu_allow_growth=gpu_allow_growth,
    ckpt_path=ckpt_path,
    log_dir=log_dir,
)

the config file I use

train:
  model:
    method: "ddf" # ddf / dvf / conditional
    backbone: "unet" # options include "local", "unet" and "global" - use "global" when method=="affine"
    local:
      num_channel_initial: 4 # 4
      extract_levels: [0, 1, 2, 3] # [0, 1, 2, 3]
    unet:
      num_channel_initial: 16
      depth: 3 # original 3
      pooling: true
      concat_skip: true

  loss:
    dissimilarity:
      image:
        name: "gmi"  # "lncc" (local normalised cross correlation), "ssd" (sum of squared distance) and "gmi" (differentiable global mutual information loss via Parzen windowing method)
        weight: 1.0
      label:
        weight: 0.0
        name: "multi_scale"
        multi_scale:
          loss_type: "dice" # options include "dice", "cross-entropy", "mean-squared", "generalised_dice" and "jaccard"
          loss_scales: [0, 1, 2, 4, 8, 16]
    regularization:
      weight: 1.0
      energy_type: "gradient-l2" # "bending" # options include "bending", "gradient-l1" and "gradient-l2"

  preprocess:
    batch_size: 2 #original 2
    shuffle_buffer_num_batch: 1

  optimizer:
    name: "adam"
    adam:
      learning_rate: 1.0e-5 #1.0e-2 bad
    sgd:
      learning_rate: 1.0e-4
      momentum: 0.9
    rms:
      learning_rate: 1.0e-4
      momentum: 0.9

  epochs: 10000
  save_period: 500