ramalama-inspect(1)

💖 ramalama-inspect — like, the manual page, bestie 💖

okay so this is ramalama-inspect(1) from the General Commands Manual, and we're gonna get into it 💅

---

✨ NAME ✨

ramalama-inspect — inspect the specified AI Model 🔍

---

🌸 SYNOPSIS 🌸

ramalama inspect [options] [model]

---

🦋 DESCRIPTION 🦋

So basically you point ramalama inspect at an AI Model and it spills the tea ☕ — extra info about the model, like its repository, its metadata, and its tensor information. It's giving detective energy, no cap 🔎✨

---

🎀 OPTIONS 🎀

• --all — prints every single thing there is to know about the AI Model. By default tho, only a basic subset gets printed. so like, --all is the full lore drop 💅
• --get=field — prints the value of one specific metadata field of the AI Model. This one supports autocomplete with the available metadata fields of the given model (iykyk, it's so handy 🫶). The special value all will print all available metadata fields and values. yes, all is special, it's the main character 🌟
• --help, -h — print usage message. the classic. iconic. never misses 💖
• --json — prints the AI Model information in json format 🧾✨

---

🌈 EXAMPLES 🌈

Inspect the smollm:135m model for basic information:

$ ramalama inspect smollm:135m
smollm:135m
   Path: /var/lib/ramalama/models/ollama/smollm:135m
   Registry: ollama
   Format: GGUF
   Version: 3
   Endianness: little
   Metadata: 39 entries
   Tensors: 272 entries

Inspect the smollm:135m model for all information in json format:

$ ramalama inspect smollm:135m --all --json
{
    "Name": "smollm:135m",
    "Path": "/home/mengel/.local/share/ramalama/models/ollama/smollm:135m",
    "Registry": "ollama",
    "Format": "GGUF",
    "Version": 3,
    "LittleEndian": true,
    "Metadata": {
        "general.architecture": "llama",
        "general.base_model.0.name": "SmolLM 135M",
        "general.base_model.0.organization": "HuggingFaceTB",
        "general.base_model.0.repo_url": "https://huggingface.co/HuggingFaceTB/SmolLM-135M",
        ...
    },
    "Tensors": [
        {
            "dimensions": [
                576,
                49152
            ],
            "n_dimensions": 2,
            "name": "token_embd.weight",
            "offset": 0,
            "type": 8
        },
        ...
    ]
}

Use the autocomplete function of --get to view a list of fields:

$ ramalama inspect smollm:135m --get general.
general.architecture               general.languages
general.base_model.0.name          general.license
general.base_model.0.organization  general.name
general.base_model.0.repo_url      general.organization
general.base_model.count           general.quantization_version
general.basename                   general.size_label
general.datasets                   general.tags
general.file_type                  general.type
general.finetune

Print the value of a specific field of the smollm:135m model:

$ ramalama inspect smollm:135m --get tokenizer.chat_template
{% for message in messages %}{{'<|im_start|>' + message['role'] + '
' + message['content'] + '<|im_end|>' + '
'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
' }}{% endif %}

Print all key-value pairs of the metadata of the smollm:135m model:

$ ramalama inspect smollm:135m --get all
general.architecture: llama
general.base_model.0.name: SmolLM 135M
general.base_model.0.organization: HuggingFaceTB
general.base_model.0.repo_url: https://huggingface.co/HuggingFaceTB/SmolLM-135M
general.base_model.count: 1
...

---

👀 SEE ALSO 👀

• ramalama(1)

go read that one next bestie, it's the whole vibe 🫶

---

📜 HISTORY 📜

Feb 2025 — originally compiled by Michael Engel mengel@redhat.com ⟨mailto:mengel@redhat.com⟩ ✨

---

and that's the whole page, fr fr 💖 this was ramalama-inspect(1) from the General Commands Manual, and you've officially read it top to bottom. slay. ate. no crumbs. periodt 💅🎀🦋