最近，我在服務器上起基于PyTorch分布式框架的預訓練實驗，起初實驗都在順利進行，但是當我們把模型的深度與寬度調大之后，模型在訓練幾代之后便會出現如下的報錯：

WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41495 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41497 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41498 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41500 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41502 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41504 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 41506 closing signal SIGTERM
ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: 1) local_rank: 1 (pid: 41496)
of binary: /home/user/anaconda3/envs/conda-envs/bin/python
Traceback (most recent call last):File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/runpy.py", line 194, in _run_module_as_mainreturn _run_code(code, main_globals, None,File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/runpy.py", line 87, in _run_codeexec(code, run_globals)File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/site-packages/torch/distributed/launch.py", l
ine 193, in <module>main()File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/site-packages/torch/distributed/launch.py", l
ine 189, in mainlaunch(args)File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/site-packages/torch/distributed/launch.py", l
ine 174, in launchrun(args)File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/site-packages/torch/distributed/run.py", line710, in runelastic_launch(File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/site-packages/torch/distributed/launcher/api.
py", line 131, in __call__
return launch_agent(self._config, self._entrypoint, list(args))
File "/home/user/anaconda3/envs/conda-envs/lib/python3.8/site-packages/torch/distributed/launcher/api.
py", line 259, in launch_agentraise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
============================================================
run_pretraining.py FAILED
------------------------------------------------------------
Failures:<NO_OTHER_FAILURES>
------------------------------------------------------------
Root Cause (first observed failure):
[0]:time      : 2024-08-30_09:05:52host      : ae83085e5bc2rank      : 1 (local_rank: 1)exitcode  : 1 (pid: 41496)error_file: <N/A>traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
============================================================

起初，我認為是batch size太大的問題，導致GPU顯存不夠，但是當我調小之后，問題照常發生。之后，我更新了PyTorch框架到2.0，但還是出現這樣的問題。
?

后續，在觀察實驗日志的時候發現，訓練期間我的梯度范數（grad_norm）變化非常不穩定，于是我順著這條線去查，遂把原因歸結為優化方面的問題。

之后，我發現對于學習率的設置，我是使用了學習率擴張法則，我的總batch為800，遠遠大于設定的256，因此導致實際訓練中，我的初始學習率由我設置的3e-4轉變為1e-3，從而導致學習率太大，進而造成了訓練坍塌。

基于上述結論，我將初始學習率調整為2e-4，模型恢復正常訓練。

上述bug出現的原因各不相同，我把我的報錯原因分享給大家，僅供參考。