Update README.md

README.md

@@ -2,7 +2,7 @@
 
 **FlagOS** is a unified heterogeneous computing software stack for large models, co-developed with leading global chip manufacturers. With core technologies such as the **FlagScale** distributed training/inference framework, **FlagGems** universal operator library, **FlagCX** communication library, and **FlagTree** unified compiler, the **FlagRelease** platform leverages the FlagOS stack to automatically produce and release various combinations of <chip + open-source model>. This enables efficient and automated model migration across diverse chips, opening a new chapter for large model deployment and application.
 
-Based on this, the **RoboBrain2.0-7B-FlagOS-FP8Dynamic** model is adapted for the Metax chip using the FlagOS software stack, enabling:
+Based on this, the **RoboBrain2.0-7B-FP8Dynamic-FlagOS** model is adapted for the Metax chip using the FlagOS software stack, enabling:
 
 ### Integrated Deployment
 
@@ -43,9 +43,18 @@ FlagEval (Libra)** is a comprehensive evaluation system and open platform for la
 
 ## Benchmark Result
 
-| Metrics     | RoboBrain2.0-7B-H100-CUDA | RoboBrain2.0-7B-FlagOS-FP8Dynamic |
-| ----------- | ------------------------- | --------------------------------- |
-| coming soon | coming soon               | coming soon                       |
+| Metrics               | RoboBrain2.0-7B-H100-CUDA | RoboBrain2.0-7B-FP8Dynamic-FlagOS |
+| --------------------- | ------------------------- | --------------------------------- |
+| SAT                   | 75.330                    | 72.000                            |
+| all_angles_bench      | 47.700                    | 46.480                            |
+| Where2Place           | 63.590                    | 63.060                            |
+| blink_val_ev          | 56.360                    | 55.200                            |
+| robo_spatial_home_all | 54.227                    | 54.312                            |
+| egoplan_bench2        | 33.230                    | 33.310                            |
+| erqa                  | 38.750                    | 39.750                            |
+| cv_bench_test         | 85.750                    | 85.770                            |
+| embspatial_bench      | 76.320                    | 75.270                            |
+| vsi_bench_tiny        | 36.100                    | 38.700                            |
 
 # User Guide
 
@@ -53,10 +62,10 @@ FlagEval (Libra)** is a comprehensive evaluation system and open platform for la
 
 **Basic Information**
 
-| Type            | Location    |
-| --------------- | ----------- |
-| Model Weights   | (https://huggingface.co/FlagRelease/RoboBrain2.0-7B-FlagOS-FP8Dynamic/tree/main) |
-| Container Image | coming soon |
+| Type            | Location                                                     |
+| --------------- | ------------------------------------------------------------ |
+| Model Weights   | https://huggingface.co/FlagRelease/RoboBrain2.0-7B-FP8Dynamic-FlagOS/files |
+| Container Image | flagrelease-registry.cn-beijing.cr.aliyuncs.com/flagrelease/flagrelease:flagrelease_nv_robobrain2_32b |
 
 **Environment Setup**
 
@@ -65,7 +74,7 @@ FlagEval (Libra)** is a comprehensive evaluation system and open platform for la
 | Accelerator Card Driver Version | Driver Version: 535.183.06            |
 | Docker Version                  | Docker version 20.10.5, build 55c4c88 |
 | Operating System                | Description:     Ubuntu 22.04.4 LTS   |
-| FlagScale                       | Version: 0.6.0                        |
+| FlagScale                       | Version: 0.8.0                        |
 | FlagGems                        | Version: 2.2                          |
 
 ## Operation Steps
@@ -74,13 +83,13 @@ FlagEval (Libra)** is a comprehensive evaluation system and open platform for la
 
 ```python
 pip install modelscope
-modelscope download --model <model path> --local_dir /share/RoboBrain2.0-7B
+modelscope download --model FlagRelease/RoboBrain2.0-7B-FP8Dynamic-FlagOS --local_dir /share/RoboBrain2.0-7B-FP8Dynamic
 ```
 
 ### Download FlagOS Image
 
 ```python
-docker pull <image>
+docker pull flagrelease-registry.cn-beijing.cr.aliyuncs.com/flagrelease/flagrelease:flagrelease_nv_robobrain2_32b
 ```
 
 ### Start the inference service
@@ -98,12 +107,25 @@ docker run --rm --init --detach \
   -v /share:/share \
   --gpus all \
   --name flagos \
-  <image> \
+  flagrelease-registry.cn-beijing.cr.aliyuncs.com/flagrelease/flagrelease:flagrelease_nv_robobrain2_32b \
   sleep infinity
   
 docker exec -it flagos bash
 ```
 
+### **Modify configuration files**
+
+```
+#Use 'pip show flag_scale' to find the installation path of FlagScale.
+pip show flag_scale
+# Modify the 7b.yaml file located at flag_scale/examples/robobrain2/conf/serve
+set the 【model path】 to /share/RoboBrain2.0-7B-FP8Dynamic
+set the 【tensor_parallel_size】 to 4
+set the 【served-model-name】 to RoboBrain2-7B-nvidia-flagos-FP8Dynamic
+# Modify the serve.yaml file located at flag_scale/examples/robobrain2/conf
+Change all the 32b to 7b in it.
+```
+
 ### Serve
 
 ```python
@@ -123,7 +145,7 @@ flagscale serve robobrain2
 import openai
 openai.api_key = "EMPTY"
 openai.base_url = "http://<server_ip>:9010/v1/"
-model = "RoboBrain2.0-7B-nv-flagos-FP8Dynamic"
+model = "RoboBrain2-7B-nvidia-flagos-FP8Dynamic"
 messages = [
     {"role": "system", "content": "You are a helpful assistant."},
     {"role": "user", "content": "What's the weather like today?"}
@@ -192,4 +214,4 @@ We warmly welcome global developers to join us:
 
 This project and related model weights are licensed under the MIT License.
 
-Release Date: 2025.07.12
+Release Date: 2025.07.15