Instructions to use FINAL-Bench/Darwin-28B-REASON with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FINAL-Bench/Darwin-28B-REASON with Transformers:

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

pipe = pipeline("text-generation", model="FINAL-Bench/Darwin-28B-REASON")
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("FINAL-Bench/Darwin-28B-REASON")
model = AutoModelForMultimodalLM.from_pretrained("FINAL-Bench/Darwin-28B-REASON", 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 FINAL-Bench/Darwin-28B-REASON with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FINAL-Bench/Darwin-28B-REASON"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/Darwin-28B-REASON",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FINAL-Bench/Darwin-28B-REASON

SGLang

How to use FINAL-Bench/Darwin-28B-REASON 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 "FINAL-Bench/Darwin-28B-REASON" \
    --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": "FINAL-Bench/Darwin-28B-REASON",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "FINAL-Bench/Darwin-28B-REASON" \
        --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": "FINAL-Bench/Darwin-28B-REASON",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use FINAL-Bench/Darwin-28B-REASON with Docker Model Runner:
```
docker model run hf.co/FINAL-Bench/Darwin-28B-REASON
```

Darwin-28B-REASON

Commit History

Jul 23

04:01
Upload README.md with huggingface_hub

SeaWolf-AI

Verified signature
af12f02
03:32
Upload README.md with huggingface_hub

SeaWolf-AI

Verified signature
d99cc14

Jun 13

10:06
Add missing preprocessor_config.json (Qwen3VLProcessor / Qwen2VLImageProcessorFast)

SeaWolf-AI

Verified signature
4669ba2

Jun 3

05:54
fix: change base_model_relation to merge for leaderboard filter compatibility

SeaWolf-AI

Verified signature
ff65d90
05:53
fix: remove incorrect datasets field (not trained on GPQA, only evaluated)

SeaWolf-AI

Verified signature
74ff480
05:44
feat: add .eval_results/gpqa_diamond.yaml for GPQA dataset indexing

SeaWolf-AI

Verified signature
239964b
05:40
fix: add datasets: Idavidrein/gpqa for dataset page indexing

SeaWolf-AI

Verified signature
2ca9efa
05:37
fix: correct model-index YAML for Idavidrein/gpqa indexing

SeaWolf-AI

Verified signature
f3b5272

May 17

09:32
Align tokenizer.json to Darwin-28B-Opus (proven-servable, identical Qwen tokenizer)

SeaWolf-AI

Verified signature
2051588
09:32
Fix tokenizer_config.json: restore chat_template + added_tokens_decoder (align to servable Opus base)

SeaWolf-AI

Verified signature
384387c
09:32
Model card: remove adapter/PEFT framing -> full standalone model (AutoModel usage, base_model_relation=finetune)

SeaWolf-AI

Verified signature
493a5af
09:24
Fix config.json: restore Qwen3_5ForConditionalGeneration + vision_config (was collapsed to text-only, mismatched model.language_model.* weights); language_model_only=true, mtp=0 to match LM-only weight set

SeaWolf-AI

Verified signature
dcad8c4
09:18
Restore model card (accidentally deleted): full Darwin-28B-REASON card, GPQA 89.39

SeaWolf-AI

Verified signature
3d8fb16
09:10
Delete adapter_config.json with huggingface_hub

SeaWolf-AI

Verified signature
c4cec9a
09:10
Delete adapter_model.safetensors with huggingface_hub

SeaWolf-AI

Verified signature
f72e22f
09:10
Darwin-28B-REASON full merged weights (base + RTD), card blank

SeaWolf-AI

Verified signature
edd7d38
08:56
Add GPQA Diamond eval result (89.39, Darwin-DELPHI)

SeaWolf-AI

Verified signature
122f43f
08:56
Finalize model card: Darwin-28B-REASON (RTD + Darwin-DELPHI, GPQA 89.39)

SeaWolf-AI

Verified signature
aa76bb8
08:52
Darwin-28B-REASON adapter weights + config (card blank)

SeaWolf-AI

Verified signature
a0cbc88
08:52
initial commit

SeaWolf-AI

Verified signature
dee71dd

Commit History

Jul 23

Upload README.md with huggingface_hub

Upload README.md with huggingface_hub

Jun 13

Add missing preprocessor_config.json (Qwen3VLProcessor / Qwen2VLImageProcessorFast)

Jun 3

fix: change base_model_relation to merge for leaderboard filter compatibility

fix: remove incorrect datasets field (not trained on GPQA, only evaluated)

feat: add .eval_results/gpqa_diamond.yaml for GPQA dataset indexing

fix: add datasets: Idavidrein/gpqa for dataset page indexing

fix: correct model-index YAML for Idavidrein/gpqa indexing

May 17

Align tokenizer.json to Darwin-28B-Opus (proven-servable, identical Qwen tokenizer)

Fix tokenizer_config.json: restore chat_template + added_tokens_decoder (align to servable Opus base)

Model card: remove adapter/PEFT framing -> full standalone model (AutoModel usage, base_model_relation=finetune)

Fix config.json: restore Qwen3_5ForConditionalGeneration + vision_config (was collapsed to text-only, mismatched model.language_model.* weights); language_model_only=true, mtp=0 to match LM-only weight set

Restore model card (accidentally deleted): full Darwin-28B-REASON card, GPQA 89.39

Delete adapter_config.json with huggingface_hub

Delete adapter_model.safetensors with huggingface_hub

Darwin-28B-REASON full merged weights (base + RTD), card blank

Add GPQA Diamond eval result (89.39, Darwin-DELPHI)

Finalize model card: Darwin-28B-REASON (RTD + Darwin-DELPHI, GPQA 89.39)

Darwin-28B-REASON adapter weights + config (card blank)

initial commit