Changes between Version 24 and Version 25 of expansion/gw16168


Ignore:
Timestamp:
08/19/2026 08:12:37 PM (2 days ago)
Author:
Tim Harvey
Comment:

updated dev kit image documentation

Legend:

Unmodified
Added
Removed
Modified

  • expansion/gw16168

    v24 v25  
    717717[https://community.nxp.com]
    718718
    719 == Development Kit Installation
    720 
    721 The development kit comes pre-installed with some initial software.
    722 
    723 Upon boot, additional software is needed as noted at boot:
    724 {{{
     719
     720[=#ai-devkit]
     721= Development Kit Installation
     722
     723The Gateworks AI development kit comes pre-installed with some initial software.
     724
     725The devkit compressed disk image is for an i.MX95 Catalina board and can be found at:
     726 - https://dev.gateworks.com/catalina/images/catalina-ai.img.gz
     727
     728Upon boot, an additional install step is required to download some python packages and models (requires Internet access):
     729 - If this one-time install needs to be run you will see:
     730{{{#!bash
    725731========================================================
    726                ARA RUNTIME INSTALL REQUIRED             
     732         Gateworks ARA AI DEVKIT IMAGE
    727733========================================================
    728 To complete the installation of the ARA2 Runtime environment please run:
     734To complete the installation of the ARA Runtime environment please run:
    729735  /root/ara2/install
    730736
    731 }}}
    732 
    733 Run the command:
    734 {{{
    735 
    736 root@catalina:~# /root/ara2/install   
     737For more instructions on working with ARA see:
     738 https://trac.gateworks.com/wiki/expansion/gw16168
     739}}}
     740
     741Example install:
     742{{{#!bash
     743root@catalina:~# /root/ara2/install
    737744
    738745╔══════════════════════════════════════════════════════════════════════════════╗
    … …  
    752759Creating new swap file of size 2G ...
    753760Setting up swapspace version 1, size = 2 GiB (2147479552 bytes)
    754 no label, UUID=23117a57-bf78-450f-afe5-69e3772faba6
     761no label, UUID=dd217f9c-d01e-44aa-91bc-ee613e7122f5
    755762/swapfile none swap sw 0 0
    756763Swap file of 2G configured and enabled successfully.
    … …  
    768775  ✓ uv package manager installed
    769776
    770 Install image-detect demo? (y/n): y
     777Installing image-detect demo...
    771778Downloading nxp/YOLOv8 to /usr/share/cnn/...
    772 Downloading bytes:                                                                                                                                                       |  0.00B            Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.                        |  0.00B /  175MB            
    773 Fetching 19 files: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:10<00:00,  1.85it/s]
    774 Download complete: : ████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████|  522MB, 87.4MB/s  Download complete: /usr/share/cnn/███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████|  746MB /  746MB,  135MB/s 
    775 Download complete: : ████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████|  522MB, 87.4MB/s 
    776 Reconstruction complete: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████|  746MB /  746MB,  135MB/s 
     779Downloading bytes:                                                                                                         |  0.00B            Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.       
     780Fetching 19 files: 100%|███████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:11<00:00,  1.62it/s]
     781Download complete: : ██████████████████████████████████████████████████████████████████████████████████████████████████████|  522MB, 90.7MB/s  Download complete: /usr/share/cnn/█████████████████████████████████████████████████████████████████████████████████|  746MB /  746MB,  119MB/s 
     782Download complete: : ██████████████████████████████████████████████████████████████████████████████████████████████████████|  522MB, 90.7MB/s 
     783Reconstruction complete: 100%|████████████████████████████████████████████████████████████████████████████████████|  746MB /  746MB,  119MB/s 
    777784Using CPython 3.14.4 interpreter at: /usr/bin/python3
    778785Creating virtual environment at: .venv
    779786Activate with: source .venv/bin/activate
    780787
    781 Install vision-webapp? (y/n): y
     788Installing vision-webapp demo...
    782789Using CPython 3.14.4 interpreter at: /usr/bin/python3
    783790Creating virtual environment at: .venv
    784791Activate with: source .venv/bin/activate
    785792
    786 Install chat Demo? (y/n): y
     793Installing chat demo...
    787794    Installing eIQ AAF Connector...
    788795
    … …  
    808815────────────────────────────────────────────────────────────────────────────────
    809816Using Python 3.14.4 environment at: /usr/share/eiq/aaf-connector/venv
    810 Resolved 56 packages in 2.01s
     817Resolved 56 packages in 1.92s
     818Downloading scipy (32.4MiB)
     819Downloading pydantic-core (1.8MiB)
     820Downloading scikit-learn (8.4MiB)
     821Downloading soundfile (1.2MiB)
     822Downloading networkx (2.0MiB)
    811823Downloading sympy (6.0MiB)
    812 Downloading networkx (2.0MiB)
    813 Downloading transformers (11.4MiB)
    814 Downloading pydantic-core (1.8MiB)
    815 Downloading av (31.8MiB)
    816 Downloading soundfile (1.2MiB)
    817 Downloading scikit-learn (8.4MiB)
    818824Downloading llvmlite (55.6MiB)
    819825Downloading hf-xet (4.1MiB)
    820 Downloading scipy (32.4MiB)
    821 Downloading pillow (6.1MiB)
     826Downloading numpy (15.0MiB)
    822827Downloading numba (3.4MiB)
    823 Downloading tokenizers (3.1MiB)
    824 Downloading numpy (15.0MiB)
     828Downloading transformers (11.4MiB)
     829Downloading av (32.1MiB)
    825830Downloading torch (99.3MiB)
    826831Downloading torchvision (2.3MiB)
     832Downloading tokenizers (3.1MiB)
     833Downloading pillow (6.1MiB)
    827834 Downloaded soundfile
    828835 Downloaded pydantic-core
    … …  
    837844 Downloaded numpy
    838845 Downloaded av
     846 Downloaded scipy
    839847 Downloaded llvmlite
    840  Downloaded scipy
    841848 Downloaded transformers
    842849 Downloaded torch
    843 Prepared 53 packages in 22.49s
    844 Installed 56 packages in 733ms
     850Prepared 53 packages in 16.77s
     851Installed 56 packages in 763ms
    845852 + annotated-types==0.8.0
    846853 + audioop-lts==0.2.2
    847854 + audioread==3.1.0
    848  + av==18.0.0
     855 + av==18.1.0
    849856 + certifi==2026.7.22
    850857 + cffi==2.1.1
    851  + charset-normalizer==3.4.9
     858 + charset-normalizer==3.5.1
    852859 + decorator==5.3.1
    853  + filelock==3.32.2
     860 + filelock==3.32.3
    854861 + fsspec==2026.7.0
    855862 + hf-xet==1.6.0
    856863 + huggingface-hub==0.36.2
    857  + idna==3.18
     864 + idna==3.19
    858865 + jinja2==3.1.6
    859866 + joblib==1.5.3
    860867 + lazy-loader==0.5
    861868 + librosa==0.11.0
    862  + llvmlite==0.48.0
     869 + llvmlite==0.49.0
    863870 + markupsafe==3.0.3
    864871 + mpmath==1.3.0
    … …  
    866873 + narwhals==2.24.0
    867874 + networkx==3.6.1
    868  + numba==0.66.0
     875 + numba==0.67.0
    869876 + numpy==2.4.3
    870877 + optimum-ara==1.0.0 (from file:///usr/share/python-wheels/optimum_ara-1.0.0-py3-none-any.whl)
    871878 + packaging==26.3
    872879 + pillow==12.1.1
    873  + platformdirs==4.11.0
     880 + platformdirs==4.11.3
    874881 + pooch==1.9.0
    875882 + pycparser==3.0
    … …  
    883890 + scikit-learn==1.9.0
    884891 + scipy==1.18.0
    885  + setuptools==83.0.0
     892 + setuptools==84.0.0
    886893 + soundfile==0.13.1
    887894 + soxr==1.1.0
    … …  
    897904 + transformers==4.57.6
    898905 + typing-extensions==4.16.0
    899  + typing-inspection==0.4.2
     906 + typing-inspection==0.4.4
    900907 + urllib3==2.7.0
    901908Using Python 3.14.4 environment at: /usr/share/eiq/aaf-connector/venv
    902 Resolved 69 packages in 737ms
     909Resolved 69 packages in 734ms
    903910Downloading opencv-python (47.5MiB)
    904911Downloading outlines-core (2.1MiB)
    905912 Downloaded outlines-core
    906913 Downloaded opencv-python
    907 Prepared 13 packages in 2.15s
     914Prepared 13 packages in 1.82s
    908915Uninstalled 1 package in 1ms
    909 Installed 14 packages in 22ms
     916Installed 14 packages in 27ms
    910917 + annotated-doc==0.0.5
    911918 + anyio==4.14.2
    … …  
    916923 + opencv-python==5.0.0.93
    917924 + outlines-core==0.2.14
    918  + pydantic-settings==2.14.2
    919  + python-dotenv==1.2.2
     925 + pydantic-settings==2.15.0
     926 + python-dotenv==1.2.3
    920927 - regex==2026.7.19
    921928 + regex==2026.2.28
    922  + sentence-transformers==5.6.1
    923  + starlette==1.4.1
    924  + uvicorn==0.52.1
     929 + sentence-transformers==5.7.0
     930 + starlette==1.6.0
     931 + uvicorn==0.52.3
    925932  ✓ Python packages installed
    926933
    … …  
    950957      - Disable auto-start: systemctl disable eiq-aaf-connector.service
    951958Downloading nxp/Qwen2.5-7B-Instruct-Ara240 to /usr/share/llm/Qwen2.5-7B-Instruct...
    952 Downloading bytes:                                                                                                                                                       |  0.00B            Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.                        |  0.00B /  0.00B            
    953 Fetching 13 files: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 13/13 [02:04<00:00,  9.56s/it]
    954 Download complete: : ████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7.78GB, 86.0MB/s  Download complete: /usr/share/llm/Qwen2.5-7B-Instruct████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8.17GB / 8.17GB, 74.4MB/s 
    955 Download complete: : ████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7.78GB, 86.0MB/s 
    956 Reconstruction complete: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8.17GB / 8.17GB, 74.4MB/s 
     959Downloading bytes:                                                                                                         |  0.00B            Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.       
     960Fetching 13 files: 100%|███████████████████████████████████████████████████████████████████████████████████████| 13/13 [02:20<00:00, 10.81s/it]
     961Download complete: : ██████████████████████████████████████████████████████████████████████████████████████████████████████| 7.78GB, 29.3MB/s  Download complete: /usr/share/llm/Qwen2.5-7B-Instruct██████████████████████████████████████████████████████████████| 8.17GB / 8.17GB, 69.3MB/s 
     962Download complete: : ██████████████████████████████████████████████████████████████████████████████████████████████████████| 7.78GB, 29.3MB/s 
     963Reconstruction complete: 100%|████████████████████████████████████████████████████████████████████████████████████| 8.17GB / 8.17GB, 69.3MB/s 
    957964Using CPython 3.14.4 interpreter at: /usr/bin/python3
    958965Creating virtual environment at: .venv
    959966Activate with: source .venv/bin/activate
    960967
    961 Install webchat Demo? (y/n): y
     968Installing webchat demo...
    962969Using CPython 3.14.4 interpreter at: /usr/bin/python3
    963970Creating virtual environment at: .venv
    964971Activate with: source .venv/bin/activate
    965972
    966 Install webVLM Demo? (y/n): y
     973Installing webVLM demo...
    967974Downloading nxp/Qwen2.5-VL-7B-Instruct-Ara240 to /usr/share/llm/Qwen2.5-VL-7B-Instruct...
    968 Downloading bytes:                                                                                                                                                       |  0.00B            Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.                        |  0.00B / 11.2GB            
    969 Fetching 19 files: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 19/19 [02:54<00:00,  9.21s/it]
    970 Download complete: : ████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 11.7GB, 35.9MB/s  Download complete: /usr/share/llm/Qwen2.5-VL-7B-Instruct█████████████████████████████████████████████████████████████████████████████████████████████████████████| 12.3GB / 12.3GB, 77.9MB/s 
    971 Download complete: : ████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 11.7GB, 35.9MB/s 
    972 Reconstruction complete: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 12.3GB / 12.3GB, 77.9MB/s 
     975Downloading bytes:                                                                                                         |  0.00B            Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.       
     976Fetching 19 files: 100%|███████████████████████████████████████████████████████████████████████████████████████| 19/19 [03:42<00:00, 11.69s/it]
     977Download complete: : ██████████████████████████████████████████████████████████████████████████████████████████████████████| 11.7GB, 64.1MB/s  Download complete: /usr/share/llm/Qwen2.5-VL-7B-Instruct███████████████████████████████████████████████████████████| 12.3GB / 12.3GB, 67.3MB/s 
     978Download complete: : ██████████████████████████████████████████████████████████████████████████████████████████████████████| 11.7GB, 64.1MB/s 
     979Reconstruction complete: 100%|████████████████████████████████████████████████████████████████████████████████████| 12.3GB / 12.3GB, 67.3MB/s 
    973980Using CPython 3.14.4 interpreter at: /usr/bin/python3
    974981Creating virtual environment at: .venv
    … …  
    981988    nxp/Qwen2.5-7B-Instruct already installed
    982989Downloading nxp/Qwen2.5-VL-7B-Instruct-Ara240 to /usr/share/llm/Qwen2.5-VL-7B-Instruct...
    983 Fetching 19 files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:00<00:00, 169.66it/s]
    984 Download complete: :                                                                                                                                                     |  0.00B            Download complete: /usr/share/llm/Qwen2.5-VL-7B-Instruct                                                                                                         |  0.00B /  0.00B           
    985 Download complete: :                                                                                                                                                     |  0.00B           
    986 Reconstruction complete: |                                                                                                                                      |  0.00B /  0.00B           
     990Fetching 19 files: 100%|██████████████████████████████████████████████████████████████████████████████████████| 19/19 [00:00<00:00, 173.71it/s]
     991Download complete: :                                                                                                       |  0.00B            Download complete: /usr/share/llm/Qwen2.5-VL-7B-Instruct                                                           |  0.00B /  0.00B           
     992Download complete: :                                                                                                       |  0.00B           
     993Reconstruction complete: |                                                                                        |  0.00B /  0.00B           
    987994
    988995
    … …  
    991998╚══════════════════════════════════════════════════════════════════════════════╝
    992999
    993 
    994 
    995 
    996 }}}
    997 
    998 Now, the demos are installed in the /root/ara2 directory.
    999 
    1000 To run a demo, use the run command in the ara2 directory shown in the example below:
    1001 {{{
    1002 root@catalina:~/# cd ara2
    1003 root@catalina:~/ara2# ./run
    1004 Usage: ./run [image-detect|vision-webapp|chat|webchat|webvlm]
    1005 root@catalina:~/ara2# ./run chat
    1006 [Config] 'Qwen2.5-7B-Instruct' is already set as enabled.
    1007 [Service] Waiting for 'Qwen2.5-7B-Instruct' to finish loading on Ara NPU (this can take 2-3 minutes)...
    1008 
    1009 [Service] AAF Connector is ready and endpoint is active!
    1010 
    1011 --- Gateworks AI LLM Session (Model: Qwen2.5-7B-Instruct) ---
    1012 Type 'exit' to stop.
    1013 
    1014 You: what is a fast animal
    1015 AI: A fast animal is one that can move very quickly, often covering long distances in a short amount of time. Some of the fastest animals include:
    1016 
    1017 1. **Cheetah (Acinonyx jubatus)**: Known for its incredible speed, the cheetah can reach up to 70-75 miles per hour (112-120 kilometers per hour) in short bursts.
    1018 
    1019 2. **Pronghorn Antelope (Antilocapra americana)**: This North American mammal can run at speeds of up to 60 miles per hour (97 kilometers per hour).
    1020 
    1021 3. **Gazelle (various species)**: Gazelles are known for their agility and can run at speeds of up to 50 miles per hour (80 kilometers per hour).
    1022 
    1023 4. **Cassowary (Casuarius casuarius)**: Although primarily a bird, the cassowary can run at speeds of up to 31 miles per hour (50 kilometers per hour) and is one of the fastest birds on land.
    1024 
    1025 5. **Cheetah's cousin, the African Wild Dog (Lycaon pictus)**: These dogs can run at speeds of up to 40-45 miles per hour (64-72 kilometers per hour).
    1026 
    1027 6. **Greyhound (Canis familiaris)**: While not as fast as the cheetah, greyhounds can reach speeds of up to 45 miles per hour (72 kilometers per hour) in short bursts.
    1028 
    1029 7. **Pronghorn Antelope's relative, the Springbok (Antidorcas marsupialis)**: This African antelope can run at speeds of up to 50 miles per hour (80 kilometers per hour).
    1030 
    1031 These animals have evolved to run fast for various reasons, such as escaping predators, catching prey, or covering large distances efficiently.
    1032 
    1033 --- Stats ---
    1034 Time taken: 91.78 seconds
    1035 Throughput: 2.59 tokens/sec
    1036 -------------
    1037 
    1038 You:
    1039 
    1040 
    1041 }}}
    1042 
    1043 All of the demos can also be seen the examples above on this page, noting the .py file will be in the respective directory.
    1044 
    1045 For example:
    1046 {{{
    1047 
    1048 root@catalina:~/# cd ara2/chat
     1000}}}
     1001
     1002
     1003Demo python applications are installed in the /root/ara2 directory:
     1004 * launcher.py: A web-based launcher for all the demos:
     1005{{{#!bash
     1006root@catalina:~# python3 /root/ara2/launcher.py
     1007
     1008=======================================================
     1009🚀 Gateworks AI Demos Launcher v1.4.0 listening on http://192.168.1.1:8888
     1010=======================================================
     1011}}}
     1012  - Now open a browser to your boards IP port 8888
     1013 * image_detect.py - Object detection on an image
     1014{{{#!bash
     1015root@catalina:~/ara2/image-detect# uv run image_detect.py
     1016Usage: image_detect.py <input_image> <output_image> [model]
     1017root@catalina:~/ara2/image-detect# uv run image_detect.py dog.jpg coco_detections.jpg /usr/share/cnn/detection/yolov8x/model.dvm
     1018
     1019model: /usr/share/cnn/detection/yolov8x/model.dvm
     1020image: dog.jpg 640x425
     1021I:DVAPP: Preprocess init: 640x640x0, size=1228800 bytes, SIMD=NEON
     1022I:DVPULB[260818164508] model type is ara2 cnn
     1023I:DVPULB[260818164509] dv_shmfd_register: register fd:  17
     1024I:DVPULB[260818164509] dv_shmfd_register: response buf_id:  127
     1025I:DVAPP: Inference init: shared memory enabled (24 MB)
     1026I:DVAPP: Inference init: 1 inputs, 2 outputs, shm=1, thread_safe=1
     1027I:DVAPP: Postprocess init: Model dimensions from input: 640x640
     1028I:DVAPP: Postprocess init: Detected YOLOv8 model - type=0, 640x640, name=noname, outputs=2
     1029DETECTIONS LOGGED: FOUND 2 ACTIVE OBJECTS
     1030----------------------------------------------------------------------
     1031Object 1: ID=13 | Name=bench | Confidence=94.1%
     1032          Bounding Box -> [113, 139] to [452, 397]
     1033----------------------------------------------------------------------
     1034Object 2: ID=16 | Name=dog | Confidence=86.7%
     1035          Bounding Box -> [208, 146] to [291, 281]
     1036----------------------------------------------------------------------
     1037SUCCESS: Mapped all boxes and text labels onto -> 'coco_detections.jpg'
     1038
     1039I:DVAPP: Preprocess cleanup: 2 frames, avg time: 843 us
     1040I:DVAPP: Inference cleanup: 2 frames, avg inference time: 75003 us (13.33 FPS)
     1041I:DVAPP: Postprocess cleanup: 2 frames, avg time: 749 us
     1042}}}
     1043 * chat.py - LLM terminal
     1044{{{#!bash
     1045root@catalina:~# cd /root/ara2/chat/
    10491046root@catalina:~/ara2/chat# uv run chat.py
    10501047[Config] 'Qwen2.5-7B-Instruct' is already set as enabled.
    10511048[Service] 'eiq-aaf-connector.service' is not active. Starting/Restarting now...
    1052 [Service] Restarting eiq-aaf-connector.service...
    1053 [Service] Waiting for 'Qwen2.5-7B-Instruct' to finish loading on Ara NPU (this can take 2-3 minutes)...
    1054 ..............................................................................
    1055 }}}
     1049[Service] Restarting eiq-aaf-connector.service... This could take several minutes
     1050[Service] Waiting for 'Qwen2.5-7B-Instruct' to finish loading on Ara NPU.....................................................................
     1051[Service] AAF Connector is ready and endpoint is active!
     1052
     1053--- Gateworks AI LLM Session (Model: Qwen2.5-7B-Instruct) ---
     1054Type 'exit' to stop.
     1055
     1056You: calculate pi to 3 places
     1057AI: To calculate π (pi) to three decimal places, you can use the approximation:
     1058
     1059\[ \pi \approx 3.142 \]
     1060
     1061This is a commonly used approximation for practical purposes.
     1062
     1063--- Stats ---
     1064Time taken: 11.33 seconds
     1065Throughput: 2.38 tokens/sec
     1066-------------
     1067}}}
     1068 * vision-webapp.py - Object detection on a video or webcam
     1069{{{#!bash
     1070root@catalina:~/ara2/vision-webapp# uv run vision-webapp.py --camera /dev/video_webcam --mp4 /usr/share/media/sample_videos/ --port 8080
     1071/root/ara2/vision-webapp/vision-webapp.py:27: DeprecationWarning: Due to '_pack_', the 'AraDetection' Structure will use memory layout compatibl
     1072e with MSVC (Windows). If this is intended, set _layout_ to 'ms'. The implicit default is deprecated and slated to become an error in Python 3.1
     10739.
     1074  class AraDetection(ctypes.Structure):
     1075Server serving on: http://localhost:8080/
     1076 * Serving Flask app 'vision-webapp'
     1077 * Debug mode: off
     1078}}}
     1079 * webchat.py - LLM Web App
     1080{{{#!bash
     1081root@catalina:~/ara2/webchat# uv run webchat.py --port 8080
     1082[Config] 'Qwen2.5-7B-Instruct' is already set as enabled.
     1083[Service] Waiting for 'Qwen2.5-7B-Instruct' to finish loading on Ara NPU...
     1084[Service] AAF Connector is ready and endpoint is active!
     1085
     1086=======================================================
     1087🚀 Gateworks Edge LLM WebChat running on http://0.0.0.0:8080
     1088=======================================================
     1089
     1090INFO:     Started server process [979]
     1091INFO:     Waiting for application startup.
     1092INFO:     Application startup complete.
     1093INFO:     Uvicorn running on http://0.0.0.0:8080 (Press CTRL+C to quit)
     1094}}}
     1095* webvlm.py - VLM Web App
     1096{{{#!bash
     1097root@catalina:~# cd /root/ara2/webvlm/
     1098root@catalina:~/ara2/webvlm# uv run webvlm.py --host 0.0.0.0 --port 8080 --video-dir /usr/share/media/sample_videos --aaf-server http://127.0.0.
     10991:8000
     1100[Config] Enabling 'Qwen2.5-VL-7B-Instruct' in server_config.json...
     1101[Service] Restarting eiq-aaf-connector.service... This could take several minutes.
     1102[Service] Waiting for 'Qwen2.5-VL-7B-Instruct' to finish loading on Ara NPU.....................................................................
     1103................................
     1104[Service] AAF Connector is ready and VLM endpoint is active!
     1105
     1106=======================================================
     1107🚀 Gateworks VLM Edge Studio running on http://0.0.0.0:8080
     1108=======================================================
     1109
     1110INFO:     Started server process [1259]
     1111INFO:     Waiting for application startup.
     1112INFO:     Application startup complete.
     1113INFO:     Uvicorn running on http://0.0.0.0:8080 (Press CTRL+C to quit)
     1114}}}