Instructions to use sionic-ai/PepperOCR-VL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sionic-ai/PepperOCR-VL with Transformers:

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

pipe = pipeline("image-text-to-text", model="sionic-ai/PepperOCR-VL")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("sionic-ai/PepperOCR-VL")
model = AutoModelForMultimodalLM.from_pretrained("sionic-ai/PepperOCR-VL", 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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use sionic-ai/PepperOCR-VL with vLLM:

Install from pip and serve model

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

SGLang

How to use sionic-ai/PepperOCR-VL 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 "sionic-ai/PepperOCR-VL" \
    --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": "sionic-ai/PepperOCR-VL",
		"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 "sionic-ai/PepperOCR-VL" \
        --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": "sionic-ai/PepperOCR-VL",
		"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 sionic-ai/PepperOCR-VL with Docker Model Runner:
```
docker model run hf.co/sionic-ai/PepperOCR-VL
```

Kindred1000 commited on 18 days ago

Commit

43fb06d

verified ·

1 Parent(s): 4f943a0

Model card: add the exact inference settings behind the reported scores

Browse files

Files changed (1) hide show

README.md +39 -4

README.md CHANGED Viewed

@@ -92,16 +92,47 @@ Message layout:
 ```
 Keep thinking mode off (`enable_thinking: false`); the model was tuned and
-evaluated without it. Decode greedily (`temperature 0`) with a mild
-`repetition_penalty` of 1.05 and `no_repeat_ngram_size` 30, and allow up to
-8192 new tokens for dense pages.
 ## Quickstart
 ### vLLM
 ```bash
-vllm serve sionic-ai/PepperOCR-VL --served-model-name pepperocr-vl --max-model-len 32768
 ```
 ```python
@@ -119,8 +150,12 @@ response = client.chat.completions.create(
         {"type": "text", "text": prompt},
     ]}],
     temperature=0.0,
     max_tokens=8192,
     extra_body={
         "repetition_penalty": 1.05,
         "no_repeat_ngram_size": 30,
         "chat_template_kwargs": {"enable_thinking": False},

 ```
 Keep thinking mode off (`enable_thinking: false`); the model was tuned and
+evaluated without it. The decoding settings are in the next section.
+## Recommended inference settings
+This is the exact configuration behind the reported scores. Use it as is
+unless you have a reason to change it.
+| Setting | Value | Note |
+|---|---|---|
+| Precision | bf16 | as shipped |
+| `temperature` | 0 | greedy |
+| `top_p` / `top_k` | 1.0 / -1 | no truncation |
+| `repetition_penalty` | 1.05 | mild; keeps long tables from looping |
+| `no_repeat_ngram_size` | 30 | blocks verbatim repetition of long spans |
+| `presence_penalty` | 0 | |
+| `max_tokens` | 8192 | enough for dense pages |
+| `seed` | 42 | for repeatability with greedy decoding |
+| `enable_thinking` | false | chat-template kwarg |
+| Image size | processor defaults | no `min_pixels`/`max_pixels` override |
+| Context (`--max-model-len`) | 32768 | |
+| Batching (`--max-num-seqs`) | 64 | with 64 concurrent requests per GPU |
+| GPU memory | `--gpu-memory-utilization 0.85` | capped at 74 GiB on larger cards |
+| Orientation preprocessing | on, threshold 0.85 | see `orientation/` |
+| Runaway-output retry | temperatures 0.1, 0.3, 0.5, 0.8 | when output hits `max_tokens` or compresses below gzip ratio 0.18; keep the shortest non-degenerate result |
+Software used for the reported scores: vLLM 0.24.0, PyTorch 2.11.0 (CUDA 13),
+Transformers 5.11.0, FlashInfer 0.6.12, `--gdn-prefill-backend triton`;
+PaddlePaddle 3.2.2 (CPU) for orientation. Newer versions should work but were
+not used for the numbers below.
 ## Quickstart
 ### vLLM
 ```bash
+vllm serve sionic-ai/PepperOCR-VL \
+  --served-model-name pepperocr-vl \
+  --max-model-len 32768 \
+  --max-num-seqs 64 \
+  --gpu-memory-utilization 0.85 \
+  --gdn-prefill-backend triton     # vLLM 0.24; drop if your version does not have it
 ```
 ```python
         {"type": "text", "text": prompt},
     ]}],
     temperature=0.0,
+    top_p=1.0,
     max_tokens=8192,
+    seed=42,
+    presence_penalty=0.0,
     extra_body={
+        "top_k": -1,
         "repetition_penalty": 1.05,
         "no_repeat_ngram_size": 30,
         "chat_template_kwargs": {"enable_thinking": False},