Instructions to use Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA with Transformers:

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

pipe = pipeline("image-text-to-text", model="Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA")
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("Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA")
model = AutoModelForMultimodalLM.from_pretrained("Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA", 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 Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA",
		"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/Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA

SGLang

How to use Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA 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 "Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA" \
    --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": "Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA",
		"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 "Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA" \
        --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": "Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA",
		"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 Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA with Docker Model Runner:
```
docker model run hf.co/Adicandra/Qwen3.5-4B-ImageCaptioning-LoRA
```

Adicandra commited on May 31

Commit

146fe1f

verified ·

1 Parent(s): 6fa9b85

Update README.md

Browse files

Files changed (1) hide show

README.md +0 -3

README.md CHANGED Viewed

@@ -37,7 +37,6 @@ Model ini merupakan hasil fine-tuning (LoRA, sudah di-**merge** ke base weights)
 | `enable_thinking` saat training | `False` |
 | EOS token | `<\|im_end\|>` (+ `<\|endoftext\|>`) |
 | Precision | bfloat16 |
-| Attention impl | FlashAttention-2 (Ampere+) / SDPA (fallback) |
 ---
@@ -57,7 +56,6 @@ model = AutoModelForImageTextToText.from_pretrained(
     REPO,
     torch_dtype=torch.bfloat16,
     device_map="auto",
-    attn_implementation="flash_attention_2",   # ganti "sdpa" jika bukan Ampere+
 )
 model.eval()
 ```
@@ -198,7 +196,6 @@ Model di-train dengan `enable_thinking=False`. Pastikan argumen tersebut selalu
 ## Keterbatasan & Catatan
 - Model hanya menghasilkan caption **Bahasa Indonesia**; tidak dirancang untuk bahasa lain.
-- Inferensi optimal pada GPU dengan dukungan **FlashAttention-2** (Ampere ke atas). Pada T4, gunakan `attn_implementation="sdpa"`.
 - Untuk input resolusi tinggi (> 560×560), model tetap berfungsi tetapi performa optimal pada thumbnail 560×560.
 - **Jangan tebak identitas/nama orang** dari gambar — sesuai aturan system prompt.
 - Evaluasi formal (CIDEr, BLEU, METEOR) belum tersedia; performa diukur secara kualitatif.

 | `enable_thinking` saat training | `False` |
 | EOS token | `<\|im_end\|>` (+ `<\|endoftext\|>`) |
 | Precision | bfloat16 |
 ---
     REPO,
     torch_dtype=torch.bfloat16,
     device_map="auto",
 )
 model.eval()
 ```
 ## Keterbatasan & Catatan
 - Model hanya menghasilkan caption **Bahasa Indonesia**; tidak dirancang untuk bahasa lain.
 - Untuk input resolusi tinggi (> 560×560), model tetap berfungsi tetapi performa optimal pada thumbnail 560×560.
 - **Jangan tebak identitas/nama orang** dari gambar — sesuai aturan system prompt.
 - Evaluasi formal (CIDEr, BLEU, METEOR) belum tersedia; performa diukur secara kualitatif.