Instructions to use cyankiwi/Inkling-Small-AWQ-INT4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cyankiwi/Inkling-Small-AWQ-INT4 with Transformers:

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

pipe = pipeline("image-text-to-text", model="cyankiwi/Inkling-Small-AWQ-INT4")
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("cyankiwi/Inkling-Small-AWQ-INT4")
model = AutoModelForMultimodalLM.from_pretrained("cyankiwi/Inkling-Small-AWQ-INT4", 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 cyankiwi/Inkling-Small-AWQ-INT4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cyankiwi/Inkling-Small-AWQ-INT4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cyankiwi/Inkling-Small-AWQ-INT4",
		"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/cyankiwi/Inkling-Small-AWQ-INT4

SGLang

How to use cyankiwi/Inkling-Small-AWQ-INT4 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 "cyankiwi/Inkling-Small-AWQ-INT4" \
    --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": "cyankiwi/Inkling-Small-AWQ-INT4",
		"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 "cyankiwi/Inkling-Small-AWQ-INT4" \
        --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": "cyankiwi/Inkling-Small-AWQ-INT4",
		"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 cyankiwi/Inkling-Small-AWQ-INT4 with Docker Model Runner:
```
docker model run hf.co/cyankiwi/Inkling-Small-AWQ-INT4
```

Run on 8x3090 (or other Ampere)

by avtc - opened about 3 hours ago

Discussion

avtc

about 3 hours ago

Good quant, thanks!

Here is a branch that adds support of Ampere cards (3090 etc).

# switch the editable install onto the mixed-precision branch
git remote add avtc https://github.com/avtc/vllm.git
git fetch avtc feature/inkling-w4a16-mixed-precision-ampere
git checkout feature/inkling-w4a16-mixed-precision-ampere

example start command:

export TORCH_CUDA_ARCH_LIST="8.6"
export CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
export RAY_memory_monitor_refresh_ms=0
export NCCL_CUMEM_ENABLE=0
export VLLM_ENABLE_CUDAGRAPH_GC=1

export VLLM_MEMORY_PROFILER_ESTIMATE_CUDAGRAPHS=0
export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
export VLLM_MARLIN_USE_ATOMIC_ADD=1
export VLLM_FLOAT32_MATMUL_PRECISION=high
export OMP_NUM_THREADS=1

vllm serve /home/ubuntu/models/Inkling-Small-AWQ-INT4-cyankiwi \
    -tp 8 \
    -dp 1 \
    --served-model-name ink-small \
    --port 8000 \
    --host 0.0.0.0 \
    --uvicorn-log-level info \
    --trust-remote-code \
    --gpu-memory-utilization 0.87 \
    --seed 1234 \
    --enable-sleep-mode \
    --skip-mm-profiling \
    -O3 \
    --no-use-tqdm-on-load \
    --tokenizer-mode inkling \
    --tool-call-parser inkling \
    --reasoning-parser inkling \
    --enable-auto-tool-choice \
    --performance-mode balanced \
    --enable-chunked-prefill \
    --max-num-seqs 4 \
    --max-model-len auto \
    --default-chat-template-kwargs '{"interleaved_thinking": true, "preserve_thinking": true}' \
    --kernel-config.enable_flashinfer_autotune=False \
    --enable-prefix-caching \
    --max-num-batched-tokens 2048

Notes:
fp16 kv-cache gives 414912 context size, vram usage 23.85GB per card. (lower --gpu-memory-utilization in case of cuda oom in the process)
fp8 or KVarN are not supported.

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