Keep saying "no CUDA", while installed Pytorch 2.0, with cuda11.7, on A100" about chatrwkv

[wensdong]

After a little back and forth, it worked. Thanks so much!

[wensdong]

The following is the full error mesaage:

"ChatRWKV project: https://github.com/BlinkDL/ChatRWKV

Loading ChatRWKV - English - cuda - fp16 - QA_PROMPT False

RWKV_JIT_ON 1

Loading model - /content/ChatRWKV/rwkv-4-pile-169m/RWKV-4-Pile-169M-20220807-8023

RuntimeError Traceback (most recent call last)
in
197
198 print(f'Loading model - {MODEL_NAME}')
--> 199 model = RWKV_RNN(args)
200
201 model_tokens = []

2 frames
/usr/local/lib/python3.8/dist-packages/torch/cuda/init.py in _lazy_init()
245 if 'CUDA_MODULE_LOADING' not in os.environ:
246 os.environ['CUDA_MODULE_LOADING'] = 'LAZY'
--> 247 torch._C._cuda_init()
248 # Some of the queued calls may reentrantly call _lazy_init();
249 # we need to just return without initializing in that case.

RuntimeError: No CUDA GPUs are available"

[BlinkDL]

pip install torch --extra-index-url https://download.pytorch.org/whl/cu117
Make sure you install torch 1.13.1+cu117