Instructions to use thinkingmachines/Inkling-Small with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use thinkingmachines/Inkling-Small with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="thinkingmachines/Inkling-Small")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("thinkingmachines/Inkling-Small")
model = AutoModelForMultimodalLM.from_pretrained("thinkingmachines/Inkling-Small", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use thinkingmachines/Inkling-Small with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "thinkingmachines/Inkling-Small"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thinkingmachines/Inkling-Small",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/thinkingmachines/Inkling-Small

SGLang

How to use thinkingmachines/Inkling-Small with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "thinkingmachines/Inkling-Small" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thinkingmachines/Inkling-Small",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "thinkingmachines/Inkling-Small" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "thinkingmachines/Inkling-Small",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use thinkingmachines/Inkling-Small with Docker Model Runner:
```
docker model run hf.co/thinkingmachines/Inkling-Small
```

QAT release?

by Downtown-Case - opened 1 day ago

Discussion

Downtown-Case

1 day ago

•

edited 1 day ago

This models seems great for its size in the playground!

I like the architecture; the 2 shared experts, in particular, are perfect for hybrid inference.

...But have you considered a QAT release for the community? Even if its just applied the expert FFNs? This would make it more accessible, and make this model stand out; currently only Gemma offers such QAT releases, and none close to this size.

CompactAI

1 day ago

I think QAT requires them to train quantized, which takes more compute. But idk.
This may be the QAT one ¯_(ツ)_/¯

kurt-tm

Thinking Machines Lab org about 24 hours ago

Please see https://huggingface.co/thinkingmachines/Inkling-Small-NVFP4

Downtown-Case

about 23 hours ago

•

edited about 23 hours ago

Thanks! I saw that, but it's just the NVFP4 quantization of the trained BF16 model, right? Or is it post-trained?

If it's just quantized, it's going to lose some intelligence compared to the BF16 weights.

I think QAT requires them to train quantized, which takes more compute. But idk.

Yeah, exactly. Ideally it would use the exact training regime of the final model (so realistically only the original model trainer can do it). It does take compute, though much less than what the final finetune should require.

I just thought it would be an interesting option, as currently there are very few large QAT models. The last "big" one to do it, that I'm aware of, is ERNIE 4.5:

https://huggingface.co/baidu/ERNIE-4.5-300B-A47B-2Bits-Paddle

They managed to squeeze it down to 2 bits with minimal loss.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment