* How to Install Ollma (Debian)
* Install ollama

        sudo apt install curl -y
        curl -fsSL https://ollama.com/install.sh | sh

* ARM64 install

        curl -L https://ollama.com/download/ollama-linux-arm64.tgz -o ollama-linux-arm64.tgz
        sudo tar -C /usr -xzf ollama-linux-arm64.tgz

* Ollama as a startup service (recommended)

        sudo useradd -r -s /bin/false -U -m -d /usr/share/ollama ollama
        sudo usermod -a -G ollama $(whoami)

* Create a service file in /etc/systemd/system/ollama.service:

        [Unit]
        Description=Ollama Service
        After=network-online.target

        [Service]
        ExecStart=/usr/bin/ollama serve
        User=ollama
        Group=ollama
        Restart=always
        RestartSec=3
        Environment="PATH=$PATH"

        [Install]
        WantedBy=default.target

* start the service

        sudo systemctl daemon-reload
        sudo systemctl enable ollama

* Start ollama

        sudo systemctl start ollama
        sudo systemctl status ollama

* Run

        ollama run deepseek-r1:1.5b

</br>

---

### 資源

* [dm-eiq-genai-flow-demonstrator](https://github.com/nxp-appcodehub/dm-eiq-genai-flow-demonstrator)
|Platform |Accelerator |Time-To-First-Token (TTFT) |Tok/s |Command|
|---|---|---|---|---|
|i.MX95 |CPU (6 threads) |0.94s |9.38 .|/eiq_genai_flow -b|
|i.MX95 |NPU (Neutron) |0.59s |9.72 |./eiq_genai_flow -b --use_neutron|

* 如何安裝
* 編譯 eIQ GenAI Flow

          $ git clone https://github.com/nxp-appcodehub/dm-eiq-genai-flow-demonstrator
          $ cp dm-eiq-genai-flow-demonstrator/meta-eiq-genai-flow <YOCTO>/sources
          $ cd <YOCTO>/sources
          $ bitbake-layers add-layer meta-eiq-genai-flow
          $ cd ../
          $ bitbake imx-image-full-eiq-genai-flow-dep

* 如何使用
* 替換dtb

            u-boot=> setenv fdtfile imx95-19x19-evk-adv7535-ap1302-neutron.dtb
            u-boot=> saveenv
            u-boot=> boot

* 下載 git

            $ git clone  https://github.com/nxp-appcodehub/dm-eiq-genai-flow-demonstrator
            $ cd ~/dm-eiq-genai-flow-demonstrator/eiq_genai_flow#

* 應用

            root@imx95-19x19-lpddr5-evk:~/dm-eiq-genai-flow-demonstrator/eiq_genai_flow# ./eiq_genai_flow -h

             Usage: eiq_genai_flow [OPTIONS]

            ╭─ Options ────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
            │ --input-mode     -i      [keyb|kasr|vasr]  The input mode: 'keyb' for Keyboard entry, 'kasr' for keyboard trigger +  │
            │                                            ASR, 'vasr' for VIT wakeWord + ASR.                                       │
            │                                            [default: keyb]                                                           │
            │ --system-prompt  -p      TEXT              System prompt for the LLM. [default: You are a helpful assistant.]        │
            │ --output-mode    -o      [text|tts]        The output mode: 'text' for textual response or 'tts' for audio response. │
            │                                            [default: tts]                                                            │
            │ --use-rag        -r                        Activate Retrieval-Augmented Generation to customize LLM outputs.         │
            │ --use-neutron    -n                        Use Neutron ONNX Execution Provider.                                      │
            │ --continuous     -c                        Continuous mode where ASR is always listening.                            │
            │ --benchmark      -b                        Benchmark mode - take a list of questions and store the results.          │
            │ --verbose        -v                        Display information (e.g. inference times).                               │
            │ --help           -h                        Show this message and exit.                                               │
            ╰──────

(1) ./eiq_genai_flow -b --use-neutron
(2) ./eiq_genai_flow --use-rag
(3) ./eiq_genai_flow --use-neutron -i keyb -o text
(4) ./eiq_genai_flow --use-neutron -i keyb -o text -r

* RAG (Data=>[medical.pdf](https://github.com/nxp-appcodehub/dm-eiq-genai-flow-demonstrator/blob/main/retrieval-database-generator/src/data/input_files/Medical.pdf))

            LLM used: danube-onnx

            Please type your question:
            How is the insulin given
            The insulin is given through a pen, syringe, or vial.

            Please type your question:
            Where do I inject the insulin?
            There are three main areas where insulin can be injected: abdomen, legs, and buttock.

* [eiq-genAI 論壇](https://community.nxp.com/t5/Generative-AI-LLMs/bd-p/Generative-AI-LLMs)
</br>

---

### 文件

* [How-to-run-deepseek-on-i-MX8MP-EVK](https://community.nxp.com/t5/i-MX-Processors-Knowledge-Base/How-to-run-deepseek-on-i-MX8MP-EVK/ta-p/2043235)
* [Run-deepseek-R1-1-5B-on-i-MX95-Android-14-0-0-2-2-0](https://community.nxp.com/t5/i-MX-Processors-Knowledge-Base/Run-deepseek-R1-1-5B-on-i-MX95-Android-14-0-0-2-2-0/ta-p/2046997?emcs_t=S2h8ZW1haWx8Ym9hcmRfc3Vic2NyaXB0aW9ufE03QkpQMkZUVUZFRUhWfDIwNDY5OTd8U1VCU0NSSVBUSU9OU3xoSw
)
* [Cross Compile / Porting the Deepseek to the i.MX 8MP|93 EVK](https://community.nxp.com/t5/i-MX-Processors-Knowledge-Base/Porting-the-Deepseek-to-the-i-MX-8MP-93-EVK/ta-p/2052745?emcs_t=S2h8ZW1haWx8Ym9hcmRfc3Vic2NyaXB0aW9ufE03TjJLNDVXNE5FQzhCfDIwNTI3NDV8U1VCU0NSSVBUSU9OU3xoSw)