Instructions to use autotrust/JEV-27B-VL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use autotrust/JEV-27B-VL with Transformers:

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

pipe = pipeline("image-text-to-text", model="autotrust/JEV-27B-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)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("autotrust/JEV-27B-VL")
model = AutoModelForMultimodalLM.from_pretrained("autotrust/JEV-27B-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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use autotrust/JEV-27B-VL with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "autotrust/JEV-27B-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": "autotrust/JEV-27B-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/autotrust/JEV-27B-VL

SGLang

How to use autotrust/JEV-27B-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 "autotrust/JEV-27B-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": "autotrust/JEV-27B-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 "autotrust/JEV-27B-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": "autotrust/JEV-27B-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 autotrust/JEV-27B-VL with Docker Model Runner:
```
docker model run hf.co/autotrust/JEV-27B-VL
```

autotrust/JEV-27B-VL

JEV-27B that can see: the same System 1 decisions and System 2, now on images

autotrust/JEV-27B-VL is autotrust/JEV-27B with vision. It pairs the multimodal Qwen3.8-27B (unmodified, including its vision encoder) with JEV-27B's System 1 adapter and decision head. One vLLM engine serves both systems, for text and for images:

	what it does	output
System 1 (`jev-decision`)	typed decisions: yes/no · pick one of 2-16 options · rate 0-5, over text and images	a calibrated probability for every option, in one forward pass
System 2 (`autotrust/JEV-27B-VL`)	the unmodified Qwen3.8-27B, optionally thinking step by step, with image input	text / reasoning

JEV-27B's language weights are bit-identical to Qwen3.8-27B's, so on text this model gives the same System 1 decisions as JEV-27B. On 24 reference decisions the largest probability difference was 0.024 (bf16 noise), with the same top option 24/24.

Highlight: #1 on the VL-RewardBench leaderboard (multimodal judge)

Given an image, a question and two answers, System 1 says which answer is better, zero-shot and in one forward pass per order. On VL-RewardBench (CVPR 2025, 1,247 human-verified pairs) it reaches 78.3% overall accuracy, the highest on the official leaderboard (26 models, updated May 2025):

judge	general	hallucination	reasoning	overall	macro
JEV-27B-VL System 1	58.0	83.2	78.2	78.3	73.1
Skywork-VL-Reward-7B	65.6	80.2	61.3	73.3	69.0
Gemini 2.0 Flash	50.8	72.6	70.1	68.8	64.5
Gemini 1.5 Pro	50.8	72.5	64.2	67.2	62.5
GPT-4o	49.1	67.6	70.5	65.8	62.4
Claude 3.5 Sonnet	43.4	55.0	62.3	55.3	53.6

2,494 decisions in 163 seconds on one GPU. Code: JEV-27B-DEMO / 09-multimodal-judge.

Highlight: zero-shot image recommendation

System 1's decision head was trained on text only. Given images, it still works zero-shot. On MicroLens-100k (real users of a short-video app, raw cover images), System 1 looked at the covers of the last 5 videos a user watched and ranked 20 candidate covers: the video the user actually watched next, hidden among 19 videos other users watched within ±3 days. 200 users.

method	uses interaction logs?	AUC	HR@5	NDCG@10
random order	no	0.489	0.205	0.219
title similarity (TF-IDF)	no	0.602	0.385	0.367
JEV-27B-VL System 1, titles only	no, zero-shot	0.649	0.455	0.399
JEV-27B-VL System 1, covers only	no, zero-shot	0.727	0.590	0.498
item-based collaborative filtering	yes, 59,045 other users	0.728	0.490	0.503

Looking beats reading: covers alone +0.078 AUC over titles (95% interval +0.031 to +0.126).
Zero-shot matches collaborative filtering learned from 59,045 users' watch histories (AUC difference 0.000, interval −0.041 to +0.040), with a higher top-5 hit rate. Brand-new videos can be ranked from their cover alone.

Code, data pipeline and a live web demo: JEV-27B-DEMO / 07-image-recommendation.

Benchmarks

JEV-27B-VL gives the same text decisions as autotrust/JEV-27B: on 1,000 answer-checking decisions the mean probability difference is 0.010 and 99.8% fall on the same side of 0.5. The text results below were measured with JEV-27B; the image result was measured with JEV-27B-VL.

Six public decision benchmarks

Scores in %, higher is better. JEV-27B and TypeSafe Jev 1.13 (hosted API) were run in full by AutoTrust (27 September 2026); the other rows are as reported in the NeoHorse-Jev-4B evaluation.

Model	JevBench	Kev	OpenJev text	Nimble	VitaminC	MASSIVE-en	Six-group mean
JEV-27B / JEV-27B-VL	88.70	83.75	73.89	92.91	77.46	87.71	84.07
TypeSafe Jev 1.13, hosted API	87.18	85.52	72.96	91.84	78.46	87.14	83.85
NeoHorse-Jev-4B	75.73	81.92	58.74	87.23	77.13	85.43	77.70
Open-Jev-9B	77.13	77.87	65.39	80.50	68.28	84.86	75.67
Kev-4B	73.71	81.47	54.75	73.40	76.46	85.71	74.25
Laya English	55.82	61.30	40.07	45.04	78.63	68.57	58.24

System 1 fidelity and calibration

Held-out test_set_30k of jev-distill-corpus-v3.

metric	JEV-27B / JEV-27B-VL
KL divergence from TypeSafe Jev 1.13's distributions (Jev-labelled rows, 0 = identical)	≈ 0.017
yes/no AUROC	0.995
choice top-1 agreement with Jev (rows with a clear top option)	95.8%
rating error, 0-5 scale (MAE of the expected rating)	0.098
expected calibration error	0.0009
KL to ground-truth labels of unseen task families	0.104

Independent benchmark: decision-models-under-pressure

Human gold labels (CLINC-150, MTOP, GoEmotions, DBpedia), 800 items.

options	2	4	8	16
TypeSafe Jev 1.13 (published)	0.890	0.801	0.782	0.769
JEV-27B / JEV-27B-VL	0.876	0.784	0.767	0.740

Applied tasks (from JEV-27B-DEMO)

task	result	comparison
Image recommendation, zero-shot (MicroLens, covers only) · JEV-27B-VL	AUC 0.727	collaborative filtering from 59,045 users' logs 0.728 · titles only 0.649
Multimodal judge (VL-RewardBench, 1,247 pairs) · JEV-27B-VL	78.3% accuracy, #1 on the leaderboard	Skywork-VL-Reward-7B 73.3 · GPT-4o 65.8 · Claude 3.5 Sonnet 55.3
Biomedical research questions (PubMedQA test, 500) · JEV-27B-VL	77.8% accuracy, zero-shot	human experts 78.0% · GPT-4 zero-shot 75.2% · BioBERT 68.1%
Response judge (RewardBench, 2,985 pairs)	89.9	Gemini 1.5 Pro 88.2 · GPT-4o 86.7 · Claude 3.5 Sonnet 84.2 as judges
Hallucination guard (TriviaQA, answer the trusted half)	96.4% accuracy	answering everything 71.2% · model's own stated confidence 87.2%
News recommendation, zero-shot (MIND)	AUC 0.642	best zero-shot baseline 0.606 · LightGBM trained on MIND 0.616
Search re-ranking (TREC-COVID, nDCG@10)	0.858	bge-reranker-v2-m3 0.793 · BM25 0.623
System 1 → System 2 (escalate below 0.70 confidence)	accuracy 0.892, 70% answered in 0.11 s	System 1 only 0.792 · thinking on everything 0.917

System 2

benchmark	result
HumanEval pass@1 (greedy)	78.0%, identical to Qwen3.8-27B

Quick start

hf download autotrust/JEV-27B-VL --local-dir JEV-27B-VL
bash JEV-27B-VL/serve.sh          # vLLM on :8000; one GPU with 80 GB or more

serve.sh runs:

vllm serve JEV-27B-VL --served-model-name autotrust/JEV-27B-VL \
  --enable-lora --max-lora-rank 32 --lora-modules jev-decision=JEV-27B-VL/adapter_vllm \
  --logprobs-mode processed_logprobs --max-model-len 16384 --enable-prefix-caching --mamba-cache-mode align \
  --limit-mm-per-prompt '{"image": 8}' --max-num-seqs 8 --trust-request-chat-template

System 1 on images

The decision prompt is sent verbatim, with images at their place in the state. The client asks only for the verbalizer tokens and applies the bundled bias and temperature:

import base64, json, math, requests

B = "JEV-27B-VL"
DH = json.load(open(f"{B}/adapter_vllm/decision_head.json"))
T = json.load(open(f"{B}/calibration.json"))["per_kind"]
RAW = ("{%- for m in messages -%}{%- for c in m['content'] -%}{%- if c['type'] == 'text' -%}{{ c['text'] }}"
       "{%- else -%}<|vision_start|><|image_pad|><|vision_end|>{%- endif -%}{%- endfor -%}{%- endfor -%}")

def image(path):
    return {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64," + base64.b64encode(open(path, "rb").read()).decode()}}

def decide_mm(kind, state_parts, question, options=None):
    """state_parts: list of str and image paths ({'image': path})."""
    options = {"noul": ["false", "true"], "score": [str(i) for i in range(6)]}.get(kind, options)
    lines = options if kind != "choice" else [f"{'ABCDEFGHIJKLMNOP'[i]}) {o}" for i, o in enumerate(options)]
    content = [{"type": "text", "text": f"[kind] {kind}\n[state] "}]
    content += [image(p["image"]) if isinstance(p, dict) else {"type": "text", "text": p} for p in state_parts]
    content.append({"type": "text", "text": f"\n[question] {question}\n[options]\n" + "\n".join(lines) + "\n[decision]:"})
    s = DH["slots"]["ranges"][kind][0]
    ids = DH["verbalizer_ids"][s: s + len(options)]
    r = requests.post("http://localhost:8000/v1/chat/completions", json={
        "model": "jev-decision", "messages": [{"role": "user", "content": content}], "chat_template": RAW,
        "add_generation_prompt": False, "add_special_tokens": False, "max_tokens": 1, "temperature": 1.0,
        "logprobs": True, "top_logprobs": len(options), "allowed_token_ids": ids, "return_tokens_as_token_ids": True}).json()
    lp = {int(t["token"].split(":")[1]): t["logprob"] for t in r["choices"][0]["logprobs"]["content"][0]["top_logprobs"]}
    z = [(lp.get(t, -1e9) + DH["bias"][s + i]) / T[kind] for i, t in enumerate(ids)]
    e = [math.exp(x - max(z)) for x in z]
    return {o: x / sum(e) for o, x in zip(options, e)}

decide_mm("noul", [{"image": "photo.jpg"}], "Is this scenario one where: the image shows food or cooking?")
# e.g. {'false': 0.00, 'true': 1.00}

Text-only System 1 decisions work exactly as with autotrust/JEV-27B (/v1/completions with the same prompt format), or through decide_mm with text parts only; both give the same result.

System 2 on images

requests.post("http://localhost:8000/v1/chat/completions", json={
    "model": "autotrust/JEV-27B-VL",
    "messages": [{"role": "user", "content": [image("chart.png"), {"type": "text", "text": "What does this chart show?"}]}],
    "max_tokens": 1024, "chat_template_kwargs": {"enable_thinking": False}})

Serving notes

--max-num-seqs 8 is required. With more than 8 sequences in one batch, vLLM's LoRA path for this multimodal model class returns wrong System 1 probabilities (the text-only JEV-27B is not affected). With the cap, results match JEV-27B at any client concurrency; requests beyond 8 simply queue.
Throughput on one B200 with the cap: 24 short text decisions in about 1.4 s; 4,000 image decisions with 6 images each in 316 s (about 13 per second).
--trust-request-chat-template lets the client send the raw decision template shown above. System 2 uses the model's own chat template.
Images are resized by the Qwen3.8 processor. Downscaling large images first (for example to at most 448 px) keeps the number of vision tokens and the latency low.

Limitations

System 1's decision head was trained on text. Decisions over images are zero-shot: they are well ordered in the experiment above, but their calibration on image tasks has not been measured systematically.
The image-recommendation result is one dataset and 200 users.

License

Apache-2.0. This repository contains the weights of Qwen/Qwen3.8-27B (Apache-2.0, see LICENSE) unchanged, plus the JEV System 1 adapter and decision head from autotrust/JEV-27B.