Instructions to use ProCreations/grug-27b-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ProCreations/grug-27b-v2 with Transformers:

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

pipe = pipeline("text-generation", model="ProCreations/grug-27b-v2")
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("ProCreations/grug-27b-v2")
model = AutoModelForMultimodalLM.from_pretrained("ProCreations/grug-27b-v2", 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 ProCreations/grug-27b-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ProCreations/grug-27b-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ProCreations/grug-27b-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ProCreations/grug-27b-v2

SGLang

How to use ProCreations/grug-27b-v2 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 "ProCreations/grug-27b-v2" \
    --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": "ProCreations/grug-27b-v2",
		"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 "ProCreations/grug-27b-v2" \
        --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": "ProCreations/grug-27b-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ProCreations/grug-27b-v2 with Docker Model Runner:
```
docker model run hf.co/ProCreations/grug-27b-v2
```

grug-27b-v2 / gguf_conversion.json

ProCreations

Release validated Grug 27B v2 with integrated MTP and measured results

f572b85 verified 19 days ago

Raw History Blame Contribute Delete

5.22 kB

	{
	"conversion_transformers": "5.16.1",
	"source_model": "ProCreations/grug-27b-v2-candidate-20260912-qwen",
	"source_revision": "79a2b2b81bf6ee0123da02467cf187db04e03708",
	"llama_cpp_commit": "2a3005c23f60cb38dab70b8ea2ddbd969bcf3e87",
	"mtp_embedded_in_every_text_gguf": true,
	"artifacts": [
	{
	"file": "grug-27b-v2-Q4_K_M.gguf",
	"bytes": 16998721440,
	"sha256": "5db0549d718c0accbe5979af0a959099f7165e3b0bea7a8604c7a1bd6872363f",
	"mtp_tensor_count": 15,
	"mtp_matrix_quantization": "Q8_0",
	"smoke_stdout": "\| model \| size \| params \| backend \| threads \| test \| t/s \|\n\| ------------------------------ \| ---------: \| ---------: \| ---------- \| ------: \| --------------: \| -------------------: \|\n\| qwen35 27B Q4_K - Medium \| 15.82 GiB \| 27.32 B \| CPU \| 24 \| pp16 \| 21.50 ± 0.00 \|\n\| qwen35 27B Q4_K - Medium \| 15.82 GiB \| 27.32 B \| CPU \| 24 \| tg8 \| 3.16 ± 0.00 \|\n\nbuild: 2a3005c (1)\n",
	"smoke_stderr": ""
	},
	{
	"file": "grug-27b-v2-Q5_K_M.gguf",
	"bytes": 19682420640,
	"sha256": "bf02d822be68618f9dca224065aa47e0cbb478de8b432a0ccc672a28d06438f8",
	"mtp_tensor_count": 15,
	"mtp_matrix_quantization": "Q8_0",
	"smoke_stdout": "\| model \| size \| params \| backend \| threads \| test \| t/s \|\n\| ------------------------------ \| ---------: \| ---------: \| ---------- \| ------: \| --------------: \| -------------------: \|\n\| qwen35 27B Q5_K - Medium \| 18.32 GiB \| 27.32 B \| CPU \| 24 \| pp16 \| 13.27 ± 0.00 \|\n\| qwen35 27B Q5_K - Medium \| 18.32 GiB \| 27.32 B \| CPU \| 24 \| tg8 \| 2.81 ± 0.00 \|\n\nbuild: 2a3005c (1)\n",
	"smoke_stderr": ""
	},
	{
	"file": "grug-27b-v2-Q6_K.gguf",
	"bytes": 22533851040,
	"sha256": "c110ad0b8863c06dacd753c41c301ba3d7c4901e5776d0d80e2396b6c184fd0c",
	"mtp_tensor_count": 15,
	"mtp_matrix_quantization": "Q8_0",
	"smoke_stdout": "\| model \| size \| params \| backend \| threads \| test \| t/s \|\n\| ------------------------------ \| ---------: \| ---------: \| ---------- \| ------: \| --------------: \| -------------------: \|\n\| qwen35 27B Q6_K \| 20.98 GiB \| 27.32 B \| CPU \| 24 \| pp16 \| 15.91 ± 0.00 \|\n\| qwen35 27B Q6_K \| 20.98 GiB \| 27.32 B \| CPU \| 24 \| tg8 \| 2.53 ± 0.00 \|\n\nbuild: 2a3005c (1)\n",
	"smoke_stderr": ""
	},
	{
	"file": "grug-27b-v2-Q8_0.gguf",
	"bytes": 29047084960,
	"sha256": "01431cb3864ee4cd7ef5a9e5d73079194f6c3bca66d0cac5ce19502e2e948f44",
	"mtp_tensor_count": 15,
	"mtp_matrix_quantization": "Q8_0",
	"smoke_stdout": "\| model \| size \| params \| backend \| threads \| test \| t/s \|\n\| ------------------------------ \| ---------: \| ---------: \| ---------- \| ------: \| --------------: \| -------------------: \|\n\| qwen35 27B Q8_0 \| 27.04 GiB \| 27.32 B \| CPU \| 24 \| pp16 \| 17.03 ± 0.00 \|\n\| qwen35 27B Q8_0 \| 27.04 GiB \| 27.32 B \| CPU \| 24 \| tg8 \| 1.92 ± 0.00 \|\n\nbuild: 2a3005c (1)\n",
	"smoke_stderr": ""
	},
	{
	"file": "grug-27b-v2-Q3_K_M.gguf",
	"bytes": 13752764320,
	"sha256": "3b36b201c4c7053bea301533d2804aa41912f701c19e320bffe0cfa5e72e07c4",
	"mtp_tensor_count": 15,
	"mtp_matrix_quantization": "Q8_0",
	"smoke_stdout": "\| model \| size \| params \| backend \| threads \| test \| t/s \|\n\| ------------------------------ \| ---------: \| ---------: \| ---------- \| ------: \| --------------: \| -------------------: \|\n\| qwen35 27B Q3_K - Medium \| 12.80 GiB \| 27.32 B \| CPU \| 24 \| pp16 \| 17.12 ± 0.00 \|\n\| qwen35 27B Q3_K - Medium \| 12.80 GiB \| 27.32 B \| CPU \| 24 \| tg8 \| 4.01 ± 0.00 \|\n\nbuild: 2a3005c (1)\n",
	"smoke_stderr": ""
	},
	{
	"file": "mmproj-grug-27b-v2-F16.gguf",
	"bytes": 927606976,
	"sha256": "1335119dd8d9f9e8c4b0a65de12b7d218cfc47282c706a0c6be0dd85be209da6"
	}
	],
	"elapsed_seconds": 1267.8098618984222,
	"conversion_time_remaining_validation": "GPU generation and embedded MTP parity must pass before public release",
	"post_conversion_validation": {
	"completed": true,
	"record": "results/runtime_validation.json",
	"basic_functional_checks": "12/12 for every quantization with MTP off and on",
	"native_api_checks": "14/15 for every mode; reasoning alias is unsupported by the pinned llama-server parser",
	"token_parity": "Measured and reported per quantization; not a bit-identical decoding guarantee",
	"coding_grader": "Corrected final-answer replay on HF CPU Jobs; see results/coding_rescore.json"
	}
	}