finally?

Gemfile

@@ -3,17 +3,10 @@
 source "https://rubygems.org"
 
 # gemspec
+gem "beautiful-jekyll-theme", "6.0.1"
 gem "ffi", "= 1.16.3"
 gem 'jemoji'
 gem 'tzinfo'
 gem 'tzinfo-data', platforms: [:mingw, :mswin, :x64_mingw]
 gem 'wdm', '>= 0.1.0'
-gem 'eventmachine', '1.2.7', git: 'https://github.com/eventmachine/eventmachine.git', tag: 'v1.2.7'
-gem "jekyll-remote-theme"
-gem 'rubyzip', '2.3.0'
-gem 'sassc' #, '2.4.0'
-gem "beautiful-jekyll-theme", "6.0.1"
-gem 'html-pipeline'
-gem 'selma', '~> 0.4.12'
-gem 'zeitwerk', '~> 2.7', '>= 2.7.1'
-gem "rouge"
+gem 'eventmachine', '1.2.7', git: 'https://github.com/eventmachine/eventmachine.git', tag: 'v1.2.7'

_posts/Coding/25-02-08-DDP.md

@@ -108,28 +108,20 @@ sidebar:
 #### DDP Initialization
 
 이제 이를 바탕으로 파이썬 코드 내에서 어떻게 세팅하는지 살펴보겠습니다. 각자마다 코딩하는 스타일이 있겠지만 저는 아래와 같이 작성하곤 합니다.
-{% highlight python %}
-args.device = 'cuda:0'
-args.world_size = 1
-args.rank = 0
-args.local_rank = int(os.environ.get("LOCAL_RANK", 0))
-torch.cuda.set_device(args.local_rank)
-torch.distributed.init_process_group(backend='nccl', init_method='env://')
-args.world_size = torch.distributed.get_world_size()
-args.local_rank = torch.distributed.get_rank()
-{% endhighlight %}
+
 <figure class='highlight'>
 <pre class='language-python' data-lang="python">
-args.device = 'cuda:0'
-args.world_size = 1
-args.rank = 0
-args.local_rank = int(os.environ.get("LOCAL_RANK", 0))
+args.device = <span class='s'>'cuda:0'</span>
+args.world_size = <span class='mi'>1</span>
+args.rank = <span class='mi'>0</span>
+args.local_rank = <span class='nb'>int</span>(os.environ.get(<span class='s'>"LOCAL_RANK"</span>, 0))
 torch.cuda.set_device(args.local_rank)
-torch.distributed.init_process_group(backend='nccl', init_method='env://')
+torch.distributed.init_process_group(backend=<span class='s'>'nccl'</span>, init_method=<span class='s'>'env://'</span>)
 args.world_size = torch.distributed.get_world_size()
 args.local_rank = torch.distributed.get_rank()
 </pre>
 </figure>
+
 ---
 
 
@@ -149,18 +141,19 @@ args.local_rank = torch.distributed.get_rank()
 
 ####  Model에 DDP 적용하기
 
-다음으로는 model에 DDP를 어떻게 적용 및 학습을 하는지 살펴보겠습니다.. (매우 간단합니다!!)
+다음으로는 model에 DDP를 어떻게 적용 및 학습을 하는지 살펴보겠습니다.. (매우 간단합니다!!
 
 <figure class='highlight'>
 <pre class='language-python' data-lang="python">
-from torch.nn.parallel import DistributedDataParallel as DDP\\
-model = DDP(model,device_ids=[args.local_rank])\\
+from <span class='nn'>torch.nn.parallel</span> import DistributedDataParallel as DDP
+model = DDP(model,device_ids=[args.local_rank])
 ...
 logits = model(x)
 loss = loss_fn(logits, labels)
 loss.backward() 
 </pre>
 </figure>
+
 ---