Instructions to use endless-frontier/Fx-Work-1028 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use endless-frontier/Fx-Work-1028 with Transformers:

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

pipe = pipeline("image-text-to-text", model="endless-frontier/Fx-Work-1028")
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("endless-frontier/Fx-Work-1028")
model = AutoModelForMultimodalLM.from_pretrained("endless-frontier/Fx-Work-1028", 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 endless-frontier/Fx-Work-1028 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "endless-frontier/Fx-Work-1028"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "endless-frontier/Fx-Work-1028",
		"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/endless-frontier/Fx-Work-1028

SGLang

How to use endless-frontier/Fx-Work-1028 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 "endless-frontier/Fx-Work-1028" \
    --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": "endless-frontier/Fx-Work-1028",
		"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 "endless-frontier/Fx-Work-1028" \
        --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": "endless-frontier/Fx-Work-1028",
		"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 endless-frontier/Fx-Work-1028 with Docker Model Runner:
```
docker model run hf.co/endless-frontier/Fx-Work-1028
```

Occasional "[non-text content omitted..." output between tool calls on long visual task (new in -1028)

by davidf419 - opened about 7 hours ago

Discussion

davidf419

about 7 hours ago

I downloaded the Q6_K quant by prithivMLmods to test locally on llama.cpp (0.7.0, q8 kv).

I gave it the prompt "Draw an SVG of an anthropomorphized cat and dog drinking coffee together at the dining room table. Iterate using vision to inspect and improve the result, writing a new SVG each time."

On its 2nd iteration, between generating a png from the svg and looking at it, it generated this text:

[non-text content omitted: unsupported by DeepSeek chat endpoint]

<｜DSML｜ calls>
<｜DSML｜ invoke name="look_at">
<｜DSML｜ parameter name="path" string="true">/iterations/v2.png</｜DSML｜ parameter>
</｜DSML｜ invoke>
</｜DSML｜ calls>

Though it followed it with a correct tool call.

Over the next iterations, after reasoning and producing svgs, before either converting it to png or viewing the image, it produced several more "[non-text content omitted: unsupported by DeepSeek chat endpoint]", always replacing a full reasoning or content entry.

I reran the prompt two more times. The 2nd try it didn't occur. The 3rd try the phrase appeared 60 times, each time preceding a tool call, usually after reasoning. Though I never saw another "<｜DSML｜ calls>".

For a minimal case, I also tested the non-chat /v1/completion API, and "[non" completed to the full phrase on every try, while the original/previous FxWork model always produced different results.

curl http://localhost:9931/v1/completions \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "[non",
    "max_tokens": 50
  }'

{"choices":[{"text":"-text content omitted: unsupported by DeepSeek chat endpoint]\n\n# 3.2.1.3\n# 3.2.1.4\n# 3.2.1.5\n# 3.2.1.", ...

My llama-server command line:

llama-server -np 1 \
    --load-mode none --image-min-tokens 1024 \
    --cache-type-k q8_0 --cache-type-v q8_0 \
    --n-cpu-moe 32 -ngl all --fit on -fa on -ub 2048 \
    -m Fx-Work-1028.Q6_K.gguf \
    --mmproj Fx-Work-1028.mmproj-bf16.gguf \
    --no-mmproj-offload \
    --reasoning-budget 10000 --reasoning-budget-message " [[I've reasoned enough. Proceeding to response.]]"

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment