Training issues with my dataset

I’m getting this issue when i run the train command:onmt_train -config data/config_kisii_en.yaml

File “C:\Users\arist\anaconda3\envs\opennmt\lib\site-packages\torch\nn\functional.py”, line 2546, in layer_norm
return torch.layer_norm(input, normalized_shape, weight, bias, eps, torch.backends.cudnn.enabled)
RuntimeError: Given normalized_shape=[256], expected input with shape [*, 256], but got input of size[32, 12, 500]

I am translating between kisii and english using data from the book of Luke. I’m using verses for every line and they’re aligned well for the book of Luke. My current configuration:
save_data: data/run/example
src_vocab: data/run/kisii_en.vocab.src
tgt_vocab: data/run/kisii_en.vocab.tgt
overwrite: False

data:
corpus_1:
path_src: data/train_source_kisii.txt # 919 verses
path_tgt: data/train_target_english.txt
valid:
path_src: data/val_source_kisii.txt # 114 verses
path_tgt: data/val_target_english.txt

world_size: 1
gpu_ranks: [0] # Remove if CUDA is False

save_model: data/run/kisii_en_model
save_checkpoint_steps: 500
train_steps: 1000 # ~35 epochs, ~35 min
valid_steps: 500

encoder_type: transformer
decoder_type: transformer
enc_layers: 2
dec_layers: 2
heads: 4
hidden_size: 256
ff_size: 512
dropout: 0.3
src_embedding_size: 256
tgt_embedding_size: 256
pos_ffn_size: 256 # Explicitly set positional encoding size
src_seq_length: 150
tgt_seq_length: 150

batch_size: 32
accum_count: 2
optim: adam
learning_rate: 0.0001
warmup_steps: 500

Any help is appreciated. Thank you