Getting running time error on Mac Pro with intel chip

KimMeen / Time-LLM

[ICLR 2024] Official implementation of " 🦙 Time-LLM: Time Series Forecasting by Reprogramming Large Language Models"

https://arxiv.org/abs/2310.01728

Apache License 2.0

1.02k stars 179 forks source link

Getting running time error on Mac Pro with intel chip #94

Closed lifgit closed 1 month ago

lifgit commented 1 month ago

It reports following error This is on a Mac Pro with interl chip. Anyone run into same error?

.conda/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py", line 1302, in _new_process_group_helper raise RuntimeError("Distributed package doesn't have NCCL built in")

lifgit commented 1 month ago

My mac doesn't have a NVIDIA GPU, does it mean distributed usage has to be disabled? If so, how could we do that?

kwuking commented 1 month ago

Hi, if you are running it locally on a CPU, you need to set num_workers to 0 and the use_gpu parameter to False. You can try running it again with these settings. However, please note that since the base model uses LLaMA2, which is an LLM, it is recommended to use 8 A100 GPUs for execution.