7B-chat model works but 13B-chat model fails

[rajendrac3]

I was able to run the llama-2-7B chat model.
But faced error in the 13B chat model.

`NOTICE: Reading model from '../llama-2-13b-chat'
NOTICE: Reading tokenizer from '../tokenizer.model'

Initializing Llama 2...
Model: ../llama-2-13b-chat

initializing model parallel with size 2
initializing ddp with size 1
initializing pipeline with size 1
Process Process-2:
Traceback (most recent call last):
File "/opt/conda/envs/pytorch/lib/python3.10/multiprocessing/process.py", line 314, in _bootstrap
self.run()
File "/opt/conda/envs/pytorch/lib/python3.10/multiprocessing/process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "/home/ec2-user/dev/rajendra/llama/llama2-flask-api/api.py", line 145, in init_process
fn(rank, size, request_queue, response_queue)
File "/home/ec2-user/dev/rajendra/llama/llama2-flask-api/api.py", line 103, in run
generator = Llama.build(
File "/home/ec2-user/dev/rajendra/llama/llama/generation.py", line 91, in build
torch.cuda.set_device(local_rank)
File "/opt/conda/envs/pytorch/lib/python3.10/site-packages/torch/cuda/init.py", line 355, in set_device
torch._C._cuda_setDevice(device)
RuntimeError: CUDA error: invalid device ordinal
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.

Loaded in 7.03 seconds
`