Created
October 9, 2023 18:59
-
-
Save HamidShojanazeri/cba001faa8471d81ba3032f42b30f703 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
(pytorch_from_src_brian_pr) hamidnazeri@a100-st-p4d24xlarge-3:~/PiPPy/examples/inference/tp-llama$ torchrun --nnodes 1 --nproc_per_node 2 chat_example.py --model_args model_args.json --tokenizer_path /data/home/hamidnazeri/PiPPy/examples/inference/model/models--meta-llama--Llama-2-13b-chat/snapshots/8ebc6a0ac2e4a781c31cb4ad395b1c26c5158c76/tokenizer.model --converted_ckpt_dir converted_checkpoints | |
[2023-10-09 18:54:44,843] torch.distributed.run: [WARNING] | |
[2023-10-09 18:54:44,843] torch.distributed.run: [WARNING] ***************************************** | |
[2023-10-09 18:54:44,843] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. | |
[2023-10-09 18:54:44,843] torch.distributed.run: [WARNING] ***************************************** | |
/opt/hpcaas/.mounts/fs-5c62ddab/home/hamidnazeri/pytorch/torch/__init__.py:632: UserWarning: torch.set_default_tensor_type() is deprecated as of PyTorch 2.1, please use torch.set_default_dtype() and torch.set_default_device() as alternatives. (Triggered internally at /opt/hpcaas/.mounts/fs-5c62ddab/home/hamidnazeri/pytorch/torch/csrc/tensor/python_tensor.cpp:450.) | |
_C._set_default_tensor_type(t) | |
/opt/hpcaas/.mounts/fs-5c62ddab/home/hamidnazeri/pytorch/torch/__init__.py:632: UserWarning: torch.set_default_tensor_type() is deprecated as of PyTorch 2.1, please use torch.set_default_dtype() and torch.set_default_device() as alternatives. (Triggered internally at /opt/hpcaas/.mounts/fs-5c62ddab/home/hamidnazeri/pytorch/torch/csrc/tensor/python_tensor.cpp:450.) | |
_C._set_default_tensor_type(t) | |
time to load the model 37.00245766097214s | |
****************************************** | |
time to load the model 37.002472958061844s | |
****************************************** | |
time for preprocessing of the chat examples is 0.0005259180907160044 s | |
======================================================================= | |
time for preprocessing of the chat examples is 0.00055427395273 s | |
======================================================================= | |
[[528, 267, 13242, 29871, 3773, 2576, 832, 22874, 297, 29889]] | |
total time profiles is [22.59880348399747] | |
len batch 1 | |
total avg inference time22.59880348399747s for total tokens of 10 and tokens per second 0.442501303534992 | |
Bandwidth achieved: 11.52 GB/s | |
======================================================================= | |
User: what is the recipe of mayonnaise? | |
> Assistant: shesopol awlim inst compens in. | |
================================== | |
[[8224, 267, 29889, 29892, 1450, 520, 304, 1171, 29901, 29883]] | |
total time profiles is [22.59822693699971] | |
len batch 1 | |
total avg inference time22.59822693699971s for total tokens of 10 and tokens per second 0.4425125930400833 | |
Bandwidth achieved: 11.52 GB/s | |
======================================================================= | |
Memory used: 17.13 GB | |
User: what is the recipe of mayonnaise? | |
> Assistant: Koes.,awost toman:c | |
================================== |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment