Translation failed when reproducing with imagent dataset

[yy-victory]

Hi,
I trained a model for 150k rounds with the parameters given in the codebase, the dataset is imagnet and a classification model, but the translation failed, what is the cause?
train dataset

MODEL_FLAGS="--attention_resolutions 32,16,8 --class_cond True --diffusion_steps 1000 --image_size 256 --learn_sigma True --noise_schedule linear --num_channels 256 --num_head_channels 64 --num_res_blocks 2 --resblock_updown True --use_fp16 True --use_scale_shift_norm True"
TRAIN_FLAGS="--batch_size 1 --lr 1e-4"
python3 scripts/image_train.py --data_dir /opt/data/private/ImageNet_ILSVRC2012/train $MODEL_FLAGS $TRAIN_FLAGS

classifier train

TRAIN_FLAGS="--iterations 300000 --anneal_lr True --batch_size 256 --lr 3e-4 --save_interval 10000 --weight_decay 0.05"
CLASSIFIER_FLAGS="--image_size 256 --classifier_attention_resolutions 32,16,8 --classifier_depth 2 --classifier_width 128 --classifier_pool attention --classifier_resblock_updown True --classifier_use_scale_shift_norm True"
python3 scripts/classifier_train.py --data_dir /opt/data/private/ImageNet_ILSVRC2012/train $TRAIN_FLAGS $CLASSIFIER_FLAGS

translation
python3 scripts/imagenet_translation.py $MODEL_FLAGS --classifier_scale 1.0 --source 260 --target 284 --val_dir /opt/data/private/ImageNet_ILSVRC2012/val --model_path /opt/data/private/ddib/models/test/ILSVRCModel/resume/model150000.pt --classifier_path /opt/data/private/ddib/models/test/classifier/model299999.pt

dog to cat （Both use the trained classifier model，This proves that the classifier is successful）
Results obtained using a pre-trained model


Results from the model I trained