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

Libraries

How to use darwinagent/WIA-27B with Transformers:

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

pipe = pipeline("text-generation", model="darwinagent/WIA-27B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("darwinagent/WIA-27B")
model = AutoModelForCausalLM.from_pretrained("darwinagent/WIA-27B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use darwinagent/WIA-27B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "darwinagent/WIA-27B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "darwinagent/WIA-27B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/darwinagent/WIA-27B

SGLang

How to use darwinagent/WIA-27B 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 "darwinagent/WIA-27B" \
    --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": "darwinagent/WIA-27B",
		"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 "darwinagent/WIA-27B" \
        --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": "darwinagent/WIA-27B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use darwinagent/WIA-27B with Docker Model Runner:
```
docker model run hf.co/darwinagent/WIA-27B
```

LightningCreeper commited on 11 days ago

Commit

355d0f3

verified ·

1 Parent(s): ba94d59

Update WIA model card tables and evaluation details

Browse files

Files changed (1) hide show

README.md +15 -13

README.md CHANGED Viewed

@@ -20,23 +20,25 @@ The model is initialized from a Qwen3.5 27B backbone and trained with WIA-CPT, o
 Results below are from the WIA technical report. Scores are on a 0--100 scale, evaluated by `DeepSeek-V4-Pro`; `Avg` is sample-weighted and `MAvg` is the macro average over seven domains.
-### WIA-Bench
-| Split | Android | MCP | OS | Search | SWE | Terminal | Web | Avg | MAvg |
 |---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
-| OOD | 50.51 | 52.53 | 63.42 | 46.36 | 59.92 | 60.73 | 54.52 | **55.45** | 55.43 |
-| ID | 43.21 | 67.74 | 35.28 | 88.54 | 66.24 | 80.85 | 73.84 | **66.41** | 65.10 |
-### Stage ablation
-| Stage | OOD Avg | ID Avg |
-|---|---:|---:|
-| Qwen3.5-27B | 49.15 | 48.91 |
-| WIA-CPT-27B | 44.19 | 43.89 |
-| WIA-SFT-27B | 46.36 | 61.96 |
-| **WIA-WM-27B** | **55.45** | **66.41** |
-The 27B run is a useful reminder that intermediate stages need not improve every aggregate: CPT and SFT are below the base OOD average, while world-model RL supplies the final OOD gain of 9.09 points from WIA-SFT-27B. The final model has the highest reported ID weighted average in the WIA-WM family.
 ## Download

 Results below are from the WIA technical report. Scores are on a 0--100 scale, evaluated by `DeepSeek-V4-Pro`; `Avg` is sample-weighted and `MAvg` is the macro average over seven domains.
+### WIA-Bench OOD
+| Model | Android | MCP | OS | Search | SWE | Terminal | Web | Avg | MAvg |
 |---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
+| WIA-WM-4B | 27.19 | 44.10 | 48.40 | 36.84 | 60.06 | 55.87 | 49.27 | 46.34 | 45.96 |
+| WIA-WM-9B | 46.82 | 52.12 | 62.70 | 46.06 | 60.77 | 59.38 | 53.43 | 54.49 | 54.47 |
+| WIA-WM-27B | 50.51 | 52.53 | 63.42 | 46.36 | 59.92 | 60.73 | 54.52 | 55.45 | 55.43 |
+| WIA-WM-35B-A3B | 48.76 | 43.72 | 63.86 | 51.33 | 59.19 | 62.52 | 54.95 | 55.72 | 54.90 |
+| WIA-WM-122B-A10B | 50.59 | 40.00 | 64.71 | 58.18 | 60.46 | 64.62 | 53.59 | 57.35 | 56.02 |
+### WIA-Bench ID
+| Model | Android | MCP | OS | Search | SWE | Terminal | Web | Avg | MAvg |
+|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
+| WIA-WM-4B | 27.12 | 65.26 | 30.76 | 82.59 | 57.52 | 75.76 | 63.24 | 59.17 | 57.46 |
+| WIA-WM-9B | 41.61 | 63.89 | 32.79 | 84.31 | 60.87 | 73.96 | 68.69 | 61.99 | 60.87 |
+| WIA-WM-27B | 43.21 | 67.74 | 35.28 | 88.54 | 66.24 | 80.85 | 73.84 | 66.41 | 65.10 |
+| WIA-WM-35B-A3B | 43.49 | 67.45 | 34.39 | 89.30 | 64.18 | 76.36 | 71.86 | 65.04 | 63.86 |
+| WIA-WM-122B-A10B | 41.44 | 64.33 | 33.01 | 86.66 | 57.74 | 71.43 | 66.21 | 61.16 | 60.12 |
 ## Download