Fix curl commands in the LLaVA server README (#1427)

Signed-off-by: dmsuehir <[email protected]>

Author: dmsuehir

comps/lvms/src/integrations/dependency/llava/README.md

@@ -55,20 +55,20 @@ cd ../../../
 docker build -t opea/lvm-llava:latest --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f comps/lvms/src/integrations/dependency/llava/Dockerfile.intel_hpu .
 ```
 
-### 2.2 Start LLaVA and LVM Service
+### 2.2 Start LLaVA Service
 
 #### 2.2.1 Start LLaVA server
 
 - Xeon
 
 ```bash
-docker run -p 8399:8399 -e http_proxy=$http_proxy --ipc=host -e https_proxy=$https_proxy opea/lvm-llava:latest
+docker run -d --name llava-service -p 8399:8399 -e http_proxy=$http_proxy --ipc=host -e https_proxy=$https_proxy opea/lvm-llava:latest
 ```
 
 - Gaudi2 HPU
 
 ```bash
-docker run -p 8399:8399 --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy opea/lvm-llava:latest
+docker run -d --name llava-service -p 8399:8399 --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy opea/lvm-llava:latest
 ```
 
 #### 2.2.2 Test
@@ -82,13 +82,13 @@ docker run -p 8399:8399 --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_M
 # Use curl/python
 
 # curl with an image and a prompt
-http_proxy="" curl http://localhost:9399/v1/lvm -XPOST -d '{"image": "iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mP8/5+hnoEIwDiqkL4KAcT9GO0U4BxoAAAAAElFTkSuQmCC", "prompt":"What is this?"}' -H 'Content-Type: application/json'
+http_proxy="" curl http://localhost:8399/generate -X POST -d '{"img_b64_str": "iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mP8/5+hnoEIwDiqkL4KAcT9GO0U4BxoAAAAAElFTkSuQmCC", "prompt":"What is this?"}' -H 'Content-Type: application/json'
 
 # curl with multiple images and a prompt (Note that depending on your MAX_IMAGES value, both images may not be sent to the LLaVA model)
-http_proxy="" curl http://localhost:9399/v1/lvm -XPOST -d '{"image": ["iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mNkYPhfz0AEYBxVSF+FAP5FDvcfRYWgAAAAAElFTkSuQmCC", "iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mNk+M9Qz0AEYBxVSF+FAAhKDveksOjmAAAAAElFTkSuQmCC"], "prompt":"What is in these images?"}' -H 'Content-Type: application/json'
+http_proxy="" curl http://localhost:8399/generate -X POST -d '{"img_b64_str": ["iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mNkYPhfz0AEYBxVSF+FAP5FDvcfRYWgAAAAAElFTkSuQmCC", "iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAYAAACNMs+9AAAAFUlEQVR42mNk+M9Qz0AEYBxVSF+FAAhKDveksOjmAAAAAElFTkSuQmCC"], "prompt":"What is in these images?"}' -H 'Content-Type: application/json'
 
 # curl with a prompt only (no image)
-http_proxy="" curl http://localhost:9399/v1/lvm -XPOST -d '{"image": "", "prompt":"What is deep learning?"}' -H 'Content-Type: application/json'
+http_proxy="" curl http://localhost:8399/generate -X POST -d '{"img_b64_str": "", "prompt":"What is deep learning?"}' -H 'Content-Type: application/json'
 
 # Test
 python check_llava_server.py