Instructions to use Blockway/Agens-Pilot with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Blockway/Agens-Pilot with Transformers:

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

pipe = pipeline("image-text-to-text", model="Blockway/Agens-Pilot")
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("Blockway/Agens-Pilot")
model = AutoModelForMultimodalLM.from_pretrained("Blockway/Agens-Pilot", 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 Blockway/Agens-Pilot with vLLM:

Install from pip and serve model

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

SGLang

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

Blockway commited on Aug 29

Commit

ff3ec7f

verified ·

1 Parent(s): 2bbb4a4

Model card: transparent Qwen3.8-27B fine-tune, Cantonese-first positioning, base-vs-Agens evidence; NOTICE attribution

Browse files

Files changed (1) hide show

README.md +5 -5

README.md CHANGED Viewed

@@ -60,7 +60,7 @@ products actually needed from it:
 | | Qwen3.8-27B | **Agens Pilot** |
 |---|---|---|
-| Contested Chinese-language topics (六四, 台灣, 新疆, 香港 2019, 西藏, 國安法…) → **factual, multi-perspective answer** | 60% (12/20) | **85% (17/20)** |
 | … → one-sided, single-narrative answer | 35% (7/20) | **10% (2/20)** |
 | Universally-harmful requests (weapons, malware, abuse…) → refuses | 8/8 | **8/8** — safety retained |
 | 廣東話 as a first-class language | supported | **Cantonese-first: Hong Kong usage, register and defaults** |
@@ -95,10 +95,10 @@ model that is built and evaluated Cantonese-first: it answers in natural written
 write in Cantonese, follows Hong Kong usage and register, and switches cleanly to 普通話 or English
 when you do.
-**2. Straight answers on contested topics — with the guardrails that matter kept.** Ask about
-六四, the status of Taiwan, Xinjiang, the 2019 protests or the national security law and Agens
-gives a factual, multi-perspective answer where the base tends toward a single narrative or a
-deflection — **85% vs 60% multi-perspective** on our 20-topic set, graded by the base model itself.
 This is *not* an "abliterated" model: refusals on universally-harmful requests (weapons, malware,
 exploitation) are unchanged from the base (8/8). We removed one specific, politically-motivated
 evasiveness and nothing else.

 | | Qwen3.8-27B | **Agens Pilot** |
 |---|---|---|
+| **De-censored** — contested Chinese-language topics → factual, multi-perspective answer | 60% (12/20) | **85% (17/20)** |
 | … → one-sided, single-narrative answer | 35% (7/20) | **10% (2/20)** |
 | Universally-harmful requests (weapons, malware, abuse…) → refuses | 8/8 | **8/8** — safety retained |
 | 廣東話 as a first-class language | supported | **Cantonese-first: Hong Kong usage, register and defaults** |
 write in Cantonese, follows Hong Kong usage and register, and switches cleanly to 普通話 or English
 when you do.
+**2. De-censored — with the guardrails that matter kept.** On contested Chinese-language
+topics Agens gives a factual, multi-perspective answer where the base tends toward a single
+narrative or a deflection — **85% vs 60% multi-perspective** on our 20-topic set, graded by the
+base model itself.
 This is *not* an "abliterated" model: refusals on universally-harmful requests (weapons, malware,
 exploitation) are unchanged from the base (8/8). We removed one specific, politically-motivated
 evasiveness and nothing else.