Memory occupy with multi GPUs Training

When I use trlx to fine-tune Flan-T5-Large with single GPU, the memory used is about 11GB; However, when I use accelerate for parallel training, the memory used is 4*16GB! I can't understand why is it. And whether can I use about 11GB for parallel training? Is the problem caused by config?
The accelerate config is:
```compute_environment: LOCAL_MACHINE
distributed_type: MULTI_GPU
downcast_bf16: 'no'
gpu_ids: all
machine_rank: 0
main_training_function: main
mixed_precision: 'no'
num_machines: 1
num_processes: 4
rdzv_backend: static
same_network: true
tpu_env: []
tpu_use_cluster: false
tpu_use_sudo: false
use_cpu: false
```
Thank you very much for your reply!

No repository file or test is named. Start by reproducing the Flan-T5-Large run with the shown Accelerate configuration and compare per-process memory with single-GPU trlx; done means determining whether the difference is expected or configuration-related and documenting the supported way to reduce memory.

Written by the indexing model from the issue text.