style: code ++

Author: Zeyi-Lin

zh/guide_cloud/integration/integration-diffsynth-studio.md

@@ -64,11 +64,11 @@ modelscope download --model=AI-ModelScope/sdxl-vae-fp16-fix --local_dir models/k
 
 如果你需要离线记录，可以添加`--swanlab_mode "local"`。
 
-```bash {3,4}
+```bash
 CUDA_VISIBLE_DEVICES="0" python examples/train/kolors/train_kolors_lora.py \
 ...
---use_swanlab \
---swanlab_mode "cloud"
+--use_swanlab \  # [!code ++]
+--swanlab_mode "cloud"  # [!code ++]
 ```
 
 ## 开启训练

zh/guide_cloud/integration/integration-easyr1.md

@@ -38,7 +38,7 @@ bash examples/run_qwen2_5_7b_math_swanlab.sh
 
 当然，这里我们可以剖析一下，由于EasyR1是原始 veRL 项目的一个干净分叉，所以继承了[veRL与SwanLab的集成](/guide_cloud/integration/integration-verl.md)。所以这里我们来看`run_qwen2_5_7b_math_swanlab.sh`文件：
 
-```sh {10}
+```sh
 set -x
 
 export VLLM_ATTENTION_BACKEND=XFORMERS
@@ -48,7 +48,7 @@ MODEL_PATH=Qwen/Qwen2.5-7B-Instruct  # replace it with your local file path
 python3 -m verl.trainer.main \
     config=examples/grpo_example.yaml \
     worker.actor.model.model_path=${MODEL_PATH} \
-    trainer.logger=['console','swanlab'] \
+    trainer.logger=['console','swanlab'] \  # [!code ++]
     trainer.n_gpus_per_node=4
 ```
 

zh/guide_cloud/integration/integration-swift.md

@@ -38,8 +38,8 @@ pip install swanlab
 ```bash
 swift sft \
     ...
-    --report_to swanlab \
-    --swanlab_project swift-robot \
+    --report_to swanlab \  # [!code ++]
+    --swanlab_project swift-robot \  # [!code ++]
     ...
 ```
 
@@ -74,8 +74,8 @@ swift sft \
     --dataloader_num_workers 4 \
     --model_author swift \
     --model_name swift-robot \
-    --report_to swanlab \
-    --swanlab_project swift-robot
+    --report_to swanlab \  # [!code ++]
+    --swanlab_project swift-robot  # [!code ++]
 ```
 
 运行指令后，就可以在SwanLab看到训练过程：

zh/guide_cloud/integration/integration-verl.md

@@ -55,10 +55,11 @@ pip install -U swanlab
 
 **完整的测试命令如下：**
 
-```bash {20}
+```bash
 PYTHONUNBUFFERED=1 python3 -m verl.trainer.main_ppo \
  data.train_files=$HOME/data/gsm8k/train.parquet \
  data.val_files=$HOME/data/gsm8k/test.parquet \
+ trainer.logger=['console','swanlab'] \  # [!code ++]
  data.train_batch_size=256 \
  data.val_batch_size=1312 \
  data.max_prompt_length=512 \
@@ -75,7 +76,6 @@ PYTHONUNBUFFERED=1 python3 -m verl.trainer.main_ppo \
  critic.model.path=Qwen/Qwen2.5-0.5B-Instruct \
  critic.ppo_micro_batch_size_per_gpu=4 \
  algorithm.kl_ctrl.kl_coef=0.001 \
- trainer.logger=['console','swanlab'] \
  +trainer.val_before_train=False \
  trainer.default_hdfs_dir=null \
  trainer.n_gpus_per_node=1 \