Review 30 (good)

subtitles/zh-CN/30_supercharge-your-pytorch-training-loop-with-accelerate.srt

@@ -20,12 +20,12 @@ There are multiple setups
 
 5
 00:00:12,600 --> 00:00:14,580
-你可以在其上进行培训：
+你可以在其上进行训练：
 on which you can run your training:
 
 6
 00:00:14,580 --> 00:00:17,910
-它可能在 CPU、GPU、TPU 上，
+它可以在 CPU、GPU、TPU 上，
 it could be on CPU, GPUs, TPUs,
 
 7
@@ -35,7 +35,7 @@ distributed on one machine with several devices,
 
 8
 00:00:20,610 --> 00:00:23,220
-甚至几台机器，通常称为节点，
+甚至几台机器，通常称为节点 (node) ，
 or even several machines, often called nodes,
 
 9
@@ -45,7 +45,7 @@ each with multiple devices.
 
 10
 00:00:26,340 --> 00:00:28,200
-最重要的是，还有新的调整
+在这之上，还有新的调整
 On top of that, there are new tweaks
 
 11
@@ -55,22 +55,22 @@ to make your training faster or more efficient,
 
 12
 00:00:30,810 --> 00:00:32,763
-比如混合精度和 DeepSpeed。
+比如混合精度和 DeepSpeed 。
 like mixed precision and DeepSpeed.
 
 13
 00:00:33,840 --> 00:00:36,600
-每一个设置或训练调整
+这每一个设置或训练调整
 Each of those setups or training tweaks
 
 14
 00:00:36,600 --> 00:00:38,760
-要求你更改训练循环的代码
+都要求你更改训练循环的代码
 requires you to change the code of your training loop
 
 15
 00:00:38,760 --> 00:00:41,733
-以某种方式学习新的 API。
+以某种方式, 或者学习新的 API。
 in one way or another and to learn a new API.
 
 16
@@ -85,7 +85,7 @@ and there are several third-party libraries that can help.
 
 18
 00:00:49,590 --> 00:00:50,760
-那些问题
+它们的问题
 The problem with those
 
 19
@@ -95,12 +95,12 @@ is that they can feel like a black box
 
 20
 00:00:53,100 --> 00:00:55,320
-并且实施调整可能并不容易
+并且实现调整可能并不容易
 and that it might not be easy to implement the tweak
 
 21
 00:00:55,320 --> 00:00:56,820
-到你需要的训练循环。
+到你需要的训练循环上。
 to the training loop you need.
 
 22
@@ -110,12 +110,12 @@ Accelerate has been designed specifically
 
 23
 00:00:59,760 --> 00:01:02,790
-让你保持对训练循环的完全控制
+以让你保持对训练循环的完全控制
 to let you retain full control over your training loop
 
 24
 00:01:02,790 --> 00:01:04,833
-并尽可能不打扰。
+并尽可能不打乱。
 and be as non-intrusive as possible.
 
 25
@@ -170,7 +170,7 @@ for your specific setup.
 
 35
 00:01:31,380 --> 00:01:33,780
-然后你必须把模型发给它，
+然后你必须发给它模型
 Then you have to send it the model,
 
 36
@@ -190,7 +190,7 @@ Accelerate handles device placement,
 
 39
 00:01:42,870 --> 00:01:44,370
-所以你不需要把你的批次
+所以你不需要把你的分批
 so you don't need to put your batch
 
 40
@@ -205,7 +205,7 @@ Finally, you have to replace the loss.backward line
 
 42
 00:01:50,640 --> 00:01:54,300
-通过 accelerator.backwardloss，
+成 accelerator.backwardloss，
 by accelerator.backwardloss,
 
 43
@@ -260,32 +260,32 @@ to the accelerator.prepare method, like for training.
 
 53
 00:02:22,170 --> 00:02:23,430
-然后你可以关闭这条线
+然后你可以关闭这行
 Then you can dismiss the line
 
 54
 00:02:23,430 --> 00:02:26,160
-将批次放在适当的设备上，
+将分批放在适当的设备上，
 that places the batch on the proper device,
 
 55
 00:02:26,160 --> 00:02:27,870
-在通过你的预测之前
+在输入你的预测
 and just before passing your predictions
 
 56
 00:02:27,870 --> 00:02:31,110
-和指标的标签，使用 accelerator.gather
+和指标的标签之前，使用 accelerator.gather
 and labels to your metric, use accelerator.gather
 
 57
 00:02:31,110 --> 00:02:33,300
-收集预测
+来收集预测
 to gather together the predictions
 
 58
 00:02:33,300 --> 00:02:34,803
-和每个过程的标签。
+和每个进程的标签。
 and labels from each process.
 
 59
@@ -295,7 +295,7 @@ A distributed training script
 
 60
 00:02:37,890 --> 00:02:41,040
-必须在不同的过程中多次启动，
+必须在不同的进程中多次启动，
 has to be launched several times on different processes,
 
 61
@@ -340,7 +340,7 @@ to generate a configuration file
 
 69
 00:03:00,330 --> 00:03:02,073
-与所有相关信息，
+含所有相关信息，
 with all the relevant information,
 
 70
@@ -355,7 +355,7 @@ followed by the path to your training script.
 
 72
 00:03:08,580 --> 00:03:12,000
-在笔记本中，你可以使用笔记本启动器功能
+在 notebook 中，你可以使用 notebook 启动器函数
 In a notebook, you can use the notebook launcher function
 
 73