Instructions to use JanHutter/verifierreward with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use JanHutter/verifierreward with Transformers:

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

pipe = pipeline("image-text-to-text", model="JanHutter/verifierreward")
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("JanHutter/verifierreward")
model = AutoModelForMultimodalLM.from_pretrained("JanHutter/verifierreward", 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 JanHutter/verifierreward with vLLM:

Install from pip and serve model

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

SGLang

How to use JanHutter/verifierreward 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 "JanHutter/verifierreward" \
    --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": "JanHutter/verifierreward",
		"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 "JanHutter/verifierreward" \
        --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": "JanHutter/verifierreward",
		"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 JanHutter/verifierreward with Docker Model Runner:
```
docker model run hf.co/JanHutter/verifierreward
```

JanHutter commited on Sep 15

Commit

08cd424

verified ·

1 Parent(s): 076eca1

Replace weights with absolute-only-ce-v1 checkpoint-1365 and document loading

Browse files

Files changed (9) hide show

README.md +20 -4
model-00001-of-00008.safetensors +1 -1
model-00002-of-00008.safetensors +1 -1
model-00003-of-00008.safetensors +1 -1
model-00004-of-00008.safetensors +1 -1
model-00005-of-00008.safetensors +1 -1
model-00006-of-00008.safetensors +1 -1
model-00007-of-00008.safetensors +1 -1
model-00008-of-00008.safetensors +1 -1

README.md CHANGED Viewed

@@ -14,6 +14,12 @@ tags:
 This is a Qwen3-VL-8B-Instruct checkpoint fine-tuned as a yes/no image-text
 alignment verifier.
 ## Load from Hugging Face
 ```python
@@ -22,12 +28,15 @@ from transformers import AutoProcessor, Qwen3VLForConditionalGeneration
 model_id = "JanHutter/verifierreward"
-processor = AutoProcessor.from_pretrained(model_id)
 model = Qwen3VLForConditionalGeneration.from_pretrained(
     model_id,
     dtype=torch.bfloat16,
     device_map="auto",
-)
 ```
@@ -38,6 +47,9 @@ evaluation is the full-vocabulary probability of the next token being `yes`.
 ```python
 import torch
 instruction = """You are an AI assistant specializing in image analysis and ranking. Your task is to analyze and compare image based on how well they match the given prompt.
 The given prompt is:{prompt}. Please consider the prompt and the image to make a decision and response directly with 'yes' or 'no'."""
@@ -45,7 +57,7 @@ The given prompt is:{prompt}. Please consider the prompt and the image to make a
 messages = [{
     "role": "user",
     "content": [
-        {"type": "image", "image": "https://example.com/image.jpg"},
         {"type": "text", "text": instruction.format(prompt="a red car in snow")},
     ],
 }]
@@ -61,7 +73,11 @@ inputs = processor.apply_chat_template(
 with torch.inference_mode():
     logits = model(**inputs, use_cache=False).logits[:, -1, :].float()
-yes_id = processor.tokenizer.encode("yes", add_special_tokens=False)[0]
 reward = torch.softmax(logits, dim=-1)[:, yes_id]
 print(reward.item())
 ```

 This is a Qwen3-VL-8B-Instruct checkpoint fine-tuned as a yes/no image-text
 alignment verifier.
+Published checkpoint: `verifierreward-qwen3vl-8b-absolute-only-ce-v1/checkpoint-1365` (training step 1365). The repository root contains the complete model, tokenizer, and image/video processor files.
+Install loading dependencies: `pip install "transformers>=4.57.6,<5" accelerate torch pillow torchvision`.
+The stored weights are float32; load with `dtype=torch.bfloat16` as below to use approximately 17.5 GB for weights, plus inference memory.
 ## Load from Hugging Face
 ```python
 model_id = "JanHutter/verifierreward"
+processor = AutoProcessor.from_pretrained(
+    model_id, max_pixels=512 * 512, fix_mistral_regex=False,
+)
 model = Qwen3VLForConditionalGeneration.from_pretrained(
     model_id,
     dtype=torch.bfloat16,
     device_map="auto",
+    attn_implementation="sdpa",
+).eval()
 ```
 ```python
 import torch
+from PIL import Image
+image = Image.open("image.jpg").convert("RGB")
 instruction = """You are an AI assistant specializing in image analysis and ranking. Your task is to analyze and compare image based on how well they match the given prompt.
 The given prompt is:{prompt}. Please consider the prompt and the image to make a decision and response directly with 'yes' or 'no'."""
 messages = [{
     "role": "user",
     "content": [
+        {"type": "image", "image": image},
         {"type": "text", "text": instruction.format(prompt="a red car in snow")},
     ],
 }]
 with torch.inference_mode():
     logits = model(**inputs, use_cache=False).logits[:, -1, :].float()
+yes_ids = processor.tokenizer.encode("yes", add_special_tokens=False)
+assert len(yes_ids) == 1
+yes_id = yes_ids[0]
 reward = torch.softmax(logits, dim=-1)[:, yes_id]
 print(reward.item())
 ```
+For reproducible loading, pass the same commit SHA as `revision` to both `from_pretrained` calls.

model-00001-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:07b1ed12130603a4d5363930c67a35dbe29c1cf9e07fe6b00b80bf61d7258eb1
 size 4962686568

 version https://git-lfs.github.com/spec/v1
+oid sha256:8b40a5d5167abfad6c82fadb3ecdc579074072f5f423965ed50c063ffe764ca1
 size 4962686568

model-00002-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ee2f57f3cb6c3851fda173fd31a6e042f9b7a79584cc6e901216182ab8ff0da
 size 4832049608

 version https://git-lfs.github.com/spec/v1
+oid sha256:75c0e35574e78e2d97745621696d8745d0c93eb2bdd8a56a45b6fc729de91335
 size 4832049608

model-00003-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eb093497cdbaa15dbba0dfdf615c142530657acdb9e6b3dd458a2e55e3bdb1f0
 size 4832049624

 version https://git-lfs.github.com/spec/v1
+oid sha256:c495730077bdbdd2655184e289e34e26021c5a6a0b5dcbfe75e6ecd6df336fde
 size 4832049624

model-00004-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2ad6714a7a09d3939fb6eabe263e0362ba8c27b558883289ac9df0655f4389b8
 size 4999856656

 version https://git-lfs.github.com/spec/v1
+oid sha256:8a08c064e8652cc2221a3a2c2aa4e00f7fce0213492777313814bd7c32a3cd4a
 size 4999856656

model-00005-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a5f05066d66bc50c83e82046250a8e7c47c2ca372c82a3abc03d1053b55477c1
 size 4832049680

 version https://git-lfs.github.com/spec/v1
+oid sha256:a768df8a91f95edd6c01c4fc775e0ff21bb0c0051e084c18bab405febec284e7
 size 4832049680

model-00006-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e62d4b3a890845109b7bd18e462f87fa7cb578a26ed3dd8c3db025d20f400273
 size 4832049672

 version https://git-lfs.github.com/spec/v1
+oid sha256:f8afbfb608841cf265f317ba915832464e7be8e65100ab1986c73e3909d8ebf6
 size 4832049672

model-00007-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:08bf618b9d5757b87f9da3f77c14507a7cb8b31bb69e71ec78ba700f6321d740
 size 3288524864

 version https://git-lfs.github.com/spec/v1
+oid sha256:961f01722d9a757d5ef2f5e9cff2e9bc3c39161dd07713839abb1ffeac535857
 size 3288524864

model-00008-of-00008.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19c691298cfd78e8df4c01335c4b8125f4b1152bd23da8f6b0d60fbd86a9c231
 size 2489319552

 version https://git-lfs.github.com/spec/v1
+oid sha256:8d307ec175c2b5f06044b639489e63542a71931f00cb1c95b1ad025b39a415ad
 size 2489319552