[Docs] Update README to use FlagScale CLI commands (flagos-ai#1130)

### PR Category
Others

### PR Types
Doc

### PR Description
docs: update README examples to FlagScale CLI

Update all command examples in README from manual script calls to 
official FlagScale CLI commands for consistency.

---------

Co-authored-by: zihugithub <fbye@baai.ac.cn>

Author: zihugithub

docs/getting-started.md

@@ -129,12 +129,14 @@ Require Megatron-LM-FL env
 
 3. Start the distributed training job:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name train action=run
+    flagscale train qwen3 --config ./examples/qwen3/conf/train.yaml
+    # or
+    flagscale train qwen3 -c ./examples/qwen3/conf/train.yaml
     ```
 
 4. Stop the distributed training job:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name train action=stop
+    flagscale train qwen3 --stop
     ```
 
 
@@ -169,7 +171,9 @@ Require vLLM-FL env
 
 3. Start inference:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name inference_fl action=run
+    flagscale inference qwen3 --config ./examples/qwen3/conf/inference_fl.yaml
+    # or
+    flagscale inference qwen3 -c ./examples/qwen3/conf/inference_fl.yaml
     ```
 
 ### Serve
@@ -214,12 +218,14 @@ Require vLLM-FL env
 
 3. Start the server:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name serve action=run
+    flagscale serve qwen3 --config ./examples/qwen3/conf/serve.yaml
+    # or
+    flagscale serve qwen3 -c ./examples/qwen3/conf/serve.yaml
     ```
 
 4. Stop the server:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name serve action=stop
+    flagscale serve qwen3 --stop
     ```
 
 ### RL
@@ -272,13 +278,15 @@ Require verl-FL env
 
 4. Start rl:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name rl action=run
+    flagscale rl qwen3 --config ./examples/qwen3/conf/rl.yaml
+    # or
+    flagscale rl qwen3 -c ./examples/qwen3/conf/rl.yaml
     ```
 You can check the output in your experiment directory.
 
 5. Stop rl:
     ```sh
-    python run.py --config-path ./examples/qwen3/conf --config-name rl action=stop
+    flagscale rl qwen3 --stop
     ```
     or force to stop ray cluster.
     ```sh

examples/emu3.5/conf/README.md

@@ -77,11 +77,15 @@ generate:
 entrypoint: `./flagscale/inference/inference_emu3p5.py`
 
 ```bash
-python run.py --config-path examples/emu3.5/conf/ --config-name image_generation.yaml
+flagscale inference emu3.5 --config ./examples/emu3.5/conf/image_generation.yaml
+# or
+flagscale inference emu3.5 -c ./examples/emu3.5/conf/image_generation.yaml
 ```
 
 ```bash
-python run.py --config-path examples/emu3.5/conf/ --config-name interleaved_generation.yaml
+flagscale inference emu3.5 --config ./examples/emu3.5/conf/interleaved_generation.yaml
+# or
+flagscale inference emu3.5 -c ./examples/emu3.5/conf/interleaved_generation.yaml
 ```
 
 ## Online Serve
@@ -104,7 +108,9 @@ Note:  You can also directly run command `python emu3p5.py`, with setting the en
 
 ```bash
 # Start emu3.5 serve
-python run.py --config-path examples/emu3.5/conf/ --config-name serve_emu3p5 action=run
+flagscale serve emu3.5 --config ./examples/emu3.5/conf/serve_emu3p5.yaml
+# or
+flagscale serve emu3.5 -c ./examples/emu3.5/conf/serve_emu3p5.yaml
 ```
 
 ```bash
@@ -114,7 +120,9 @@ python client.py
 
 ```bash
 # Stop emu3.5 serve
-python run.py --config-path examples/emu3.5/conf/ --config-name serve_emu3p5 action=stop
+flagscale serve emu3.5 --config ./examples/emu3.5/conf/serve_emu3p5.yaml --stop
+# or
+flagscale serve emu3.5 -c ./examples/emu3.5/conf/serve_emu3p5.yaml --stop
 ```
 
 ```python

examples/pi0/README.md

@@ -181,7 +181,9 @@ Configure the following fields:
 ### Start Training
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0/conf --config-name train action=run
+flagscale train pi0 --config ./examples/pi0/conf/train.yaml
+# or
+flagscale train pi0 -c ./examples/pi0/conf/train.yaml
 ```
 
 Training logs are saved to `outputs/pi0_train/logs/host_0_localhost.output` by default.
@@ -191,7 +193,7 @@ Checkpoints are saved to `${experiment.exp_dir}/checkpoints` (default: `outputs/
 ### Stop Training
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0/conf --config-name train action=stop
+flagscale train pi0 --stop
 ```
 
 ## Inference
@@ -270,10 +272,9 @@ Configure the following fields:
 
 ```sh
 cd FlagScale/
-python run.py \
-    --config-path ./examples/pi0/conf \
-    --config-name inference \
-    action=run
+flagscale inference pi0 --config ./examples/pi0/conf/inference.yaml
+# or
+flagscale inference pi0 -c ./examples/pi0/conf/inference.yaml
 ```
 
 Inference logs are saved to `outputs/pi0_inference/inference_logs/host_0_localhost.output` by default.
@@ -312,7 +313,9 @@ Configure the following fields:
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0/conf --config-name serve action=run
+flagscale serve pi0 --config ./examples/pi0/conf/serve.yaml
+# or
+flagscale serve pi0 -c ./examples/pi0/conf/serve.yaml
 ```
 
 Serving logs are saved to `outputs/pi0_serve/logs/host_0_localhost.output` by default.
@@ -321,7 +324,7 @@ Serving logs are saved to `outputs/pi0_serve/logs/host_0_localhost.output` by de
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0/conf --config-name serve action=stop
+flagscale serve pi0 --stop
 ```
 
 ### Test Server with Client

examples/pi0_5/README.md

@@ -188,7 +188,9 @@ Configure the following fields:
 ### Start Training
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0_5/conf --config-name train action=run
+flagscale train pi0_5 --config ./examples/pi0_5/conf/train.yaml
+# or
+flagscale train pi0_5 -c ./examples/pi0_5/conf/train.yaml
 ```
 
 Training logs are saved to `outputs/pi0_5_train/logs/host_0_localhost.output` by default.
@@ -198,7 +200,7 @@ Checkpoints are saved to `${experiment.exp_dir}/checkpoints` (default: `outputs/
 ### Stop Training
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0_5/conf --config-name train action=stop
+flagscale train pi0_5 --stop
 ```
 
 ## Inference
@@ -279,10 +281,9 @@ Configure the following fields:
 
 ```sh
 cd FlagScale/
-python run.py \
-    --config-path ./examples/pi0_5/conf \
-    --config-name inference \
-    action=run
+flagscale inference pi0_5 --config ./examples/pi0_5/conf/inference.yaml
+# or
+flagscale inference pi0_5 -c ./examples/pi0_5/conf/inference.yaml
 ```
 
 Inference logs are saved to `outputs/pi0_5_inference/inference_logs/host_0_localhost.output` by default.
@@ -323,7 +324,9 @@ Configure the following fields:
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0_5/conf --config-name serve action=run
+flagscale serve pi0_5 --config ./examples/pi0_5/conf/serve.yaml
+# or
+flagscale serve pi0_5 -c ./examples/pi0_5/conf/serve.yaml
 ```
 
 Serving logs are saved to `outputs/pi0_5_serve/logs/host_0_localhost.output` by default.
@@ -332,7 +335,7 @@ Serving logs are saved to `outputs/pi0_5_serve/logs/host_0_localhost.output` by
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/pi0_5/conf --config-name serve action=stop
+flagscale serve pi0_5 --stop
 ```
 
 ### Test Server with Client

examples/qwen2_5_vl/README.md

@@ -155,12 +155,14 @@ tokenizer_path: /mnt/qwen2.5-vl-ckpts/Qwen2.5-VL-7B-Instruct-tp2
 
 Start training.
 ```bash
-python run.py --config-path ./examples/qwen2_5_vl/conf  --config-name train action=run
+flagscale train qwen2_5_vl --config ./examples/qwen2_5_vl/conf/train.yaml
+# or
+flagscale train qwen2_5_vl -c ./examples/qwen2_5_vl/conf/train.yaml
 ```
 
 Stop training.
 ```bash
-python run.py --config-path ./examples/qwen2_5_vl/conf  --config-name train action=stop
+flagscale train qwen2_5_vl --stop
 ```
 
 ### Convert the checkpoint to HuggingFace
@@ -186,7 +188,9 @@ More details about [Auto-Evaluation](https://github.com/flageval-baai/Auto-Evalu
 ### Start the server
 
     ```sh
-    python run.py --config-path ./examples/robobrain2/conf --config-name serve action=run
+    flagscale serve robobrain2 --config ./examples/robobrain2/conf/serve.yaml
+    # or
+    flagscale serve robobrain2 -c ./examples/robobrain2/conf/serve.yaml
     ```
 
 ### Start evaluation

examples/qwen3/README.md

@@ -98,7 +98,9 @@ Start server with FlagScale:
 
 ```sh
 cd FlagScale
-python run.py --config-path ./examples/qwen3/conf --config-name serve action=run
+flagscale serve qwen3 --config ./examples/qwen3/conf/serve.yaml
+# or
+flagscale serve qwen3 -c ./examples/qwen3/conf/serve.yaml
 ```
 
 Check logs:

examples/qwen3_vl/README.md

@@ -142,12 +142,14 @@ tokenizer_path: /mnt/qwen-vl-ckpts/Qwen3-VL-8B-Instruct
 
 Start training.
 ```bash
-python run.py --config-path ./examples/qwen3_vl/conf  --config-name train action=run
+flagscale train qwen3_vl --config ./examples/qwen3_vl/conf/train.yaml
+# or
+flagscale train qwen3_vl -c ./examples/qwen3_vl/conf/train.yaml
 ```
 
 Stop training.
 ```bash
-python run.py --config-path ./examples/qwen3_vl/conf  --config-name train action=stop
+flagscale train qwen3_vl --stop
 ```
 
 ### 5. Convert the checkpoint to HuggingFace
@@ -177,7 +179,9 @@ More details about [Auto-Evaluation](https://github.com/flageval-baai/Auto-Evalu
     More details refer to this [doc](https://github.com/flagos-ai/FlagScale/tree/main/examples/robobrain2_5).
 
     ```sh
-    python run.py --config-path ./examples/robobrain2_5/conf --config-name serve action=run
+    flagscale serve robobrain2_5 --config ./examples/robobrain2_5/conf/serve.yaml
+    # or
+    flagscale serve robobrain2_5 -c ./examples/robobrain2_5/conf/serve.yaml
     ```
 
 ### 2. Start evaluation

examples/robobrain2/README.md

@@ -70,7 +70,9 @@ Change 2 fields:
 ### Run Inference
 
 ```sh
-python run.py --config-path ./examples/robobrain2/conf --config-name inference action=run
+flagscale inference robobrain2 --config ./examples/robobrain2/conf/inference.yaml
+# or
+flagscale inference robobrain2 -c ./examples/robobrain2/conf/inference.yaml
 ```
 
 ### Check Logs
@@ -97,7 +99,9 @@ Change 1 fields:
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/robobrain2/conf --config-name serve action=run
+flagscale serve robobrain2 --config ./examples/robobrain2/conf/serve.yaml
+# or
+flagscale serve robobrain2 -c ./examples/robobrain2/conf/serve.yaml
 ```
 
 ## Test Server with CURL

examples/robobrain2_5/README.md

@@ -71,7 +71,9 @@ Change 2 fields:
 ### Run Inference
 
 ```sh
-python run.py --config-path ./examples/robobrain2_5/conf --config-name inference action=run
+flagscale inference robobrain2_5 --config ./examples/robobrain2_5/conf/inference.yaml
+# or
+flagscale inference robobrain2_5 -c ./examples/robobrain2_5/conf/inference.yaml
 ```
 
 ### Check Logs
@@ -98,7 +100,9 @@ Change 1 fields:
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/robobrain2_5/conf --config-name serve action=run
+flagscale serve robobrain2_5 --config ./examples/robobrain2_5/conf/serve.yaml
+# or
+flagscale serve robobrain2_5 -c ./examples/robobrain2_5/conf/serve.yaml
 ```
 
 ## Test Server with CURL

examples/robobrain_x0/README.md

@@ -119,7 +119,9 @@ Change 2 fields:
 ### Start Training
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/robobrain_x0/conf/ --config-name train action=run
+flagscale train robobrain_x0 --config ./examples/robobrain_x0/conf/train.yaml
+# or
+flagscale train robobrain_x0 -c ./examples/robobrain_x0/conf/train.yaml
 ```
 
 ## Training with Lerobot Dataset
@@ -217,7 +219,9 @@ defaults:
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/robobrain_x0/conf/ --config-name train action=run
+flagscale train robobrain_x0 --config ./examples/robobrain_x0/conf/train.yaml
+# or
+flagscale train robobrain_x0 -c ./examples/robobrain_x0/conf/train.yaml
 ```
 
 ### Supported Video Backends
@@ -255,7 +259,9 @@ Change 3 fields:
 
 ```sh
 cd FlagScale/
-python run.py --config-path ./examples/robobrain_x0/conf --config-name serve action=run
+flagscale serve robobrain_x0 --config ./examples/robobrain_x0/conf/serve.yaml
+# or
+flagscale serve robobrain_x0 -c ./examples/robobrain_x0/conf/serve.yaml
 ```
 
 ### Test Server with Client