Instructions to use TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ with Transformers:

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

pipe = pipeline("image-text-to-text", model="TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ")
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("TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ")
model = AutoModelForMultimodalLM.from_pretrained("TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ", 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 TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ",
		"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/TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ

SGLang

How to use TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ 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 "TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ" \
    --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": "TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ",
		"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 "TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ" \
        --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": "TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ",
		"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 TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ with Docker Model Runner:
```
docker model run hf.co/TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ
```

selimaktas commited on 3 days ago

Commit

6621bc4

verified ·

1 Parent(s): 40f6bb2

Upload README.md with huggingface_hub

Browse files

Files changed (1) hide show

README.md +10 -9

README.md CHANGED Viewed

@@ -18,7 +18,7 @@ Mixed-precision **NVFP4** quantization of [`Qwen/Qwen3.8-27B`](https://huggingfa
 built with [llm-compressor](https://github.com/vllm-project/llm-compressor) using
 **AWQ activation-aware scaling followed by GPTQ**, with an `imatrix_mse` observer.
-**24 GB.** At the same size as a plain NVFP4 build, it cuts high-confidence damage by
 roughly a third, and it is the most accurate NVFP4 checkpoint of this model we have measured.
 ## Recipe
@@ -53,14 +53,15 @@ output, plus 200 free greedy generations. vLLM 0.27.1, TP=2, 2×B300.
 | checkpoint | size ↓ | top-1 ↑ | near-tie ↓ | moderate ↓ | confident ↓ | certain ↓ | divmed ↑ | tok/s ↑ |
 |---|---:|---:|---:|---:|---:|---:|---:|---:|
-| `Qwen/Qwen3.8-27B-FP8` *(8-bit ref)* | 29 GB | 96.15% | 22.70% | 3.48% | 1.45% | 0.08% | 47 | 8711 |
-| **this model (NVFP4+AWQ)** | 24 GB | **93.44%** | **33.86%** | **7.74%** | **2.69%** | **0.19%** | **29** | 10680 |
-| `RadixArk/Qwen3.8-27B-NVFP4` | **21 GB** | 90.23% | 43.80% | 14.49% | 3.29% | 0.70% | 11 | **11436** |
-| `unsloth/Qwen3.8-27B-NVFP4` | 22 GB | 91.75% | 40.12% | 10.32% | 3.91% | 0.25% | 19 | 11069 |
 Bold marks the best value in each column **among the FP4 checkpoints**; the FP8 row is a
 reference at a different precision and size class, so it is excluded from the comparison.
-Sizes for our checkpoints include the ~0.85 GB BF16 MTP head; third-party rows do not have one.
 **Columns.** `top-1` is raw argmax agreement with BF16. The four bucket columns are
 *disagreement* rates, split by how confident the base model was at that position
@@ -100,9 +101,9 @@ llm = LLM("TelperionAI/Qwen3.8-27B-NVFP4-AWQ-GPTQ", tensor_parallel_size=2,
 ```
 `Qwen3_5ForConditionalGeneration` does not carry `mtp.*` in its state dict, so
-llm-compressor never sees it and it is absent from most community quantizations of this
-model — even though `config.json` declares `mtp_num_hidden_layers: 1`. It is grafted back
-in here from the base checkpoint and excluded from quantization (`re:.*mtp.*` in
 `quantization_config.ignore`; without that exclusion the quantization target regexes also
 match `mtp.layers.0.mlp.*` and vLLM fails to load). Draft quality drives acceptance rate,
 so it is kept at full precision rather than quantized.

 built with [llm-compressor](https://github.com/vllm-project/llm-compressor) using
 **AWQ activation-aware scaling followed by GPTQ**, with an `imatrix_mse` observer.
+**24.7 GB.** At the same size as a plain NVFP4 build, it cuts high-confidence damage by
 roughly a third, and it is the most accurate NVFP4 checkpoint of this model we have measured.
 ## Recipe
 | checkpoint | size ↓ | top-1 ↑ | near-tie ↓ | moderate ↓ | confident ↓ | certain ↓ | divmed ↑ | tok/s ↑ |
 |---|---:|---:|---:|---:|---:|---:|---:|---:|
+| `Qwen/Qwen3.8-27B-FP8` *(8-bit ref)* | 30.9 GB | 96.15% | 22.70% | 3.48% | 1.45% | 0.08% | 47 | 8711 |
+| **this model (NVFP4+AWQ)** | 24.7 GB | **93.44%** | **33.86%** | **7.74%** | **2.69%** | **0.19%** | **29** | 10680 |
+| `RadixArk/Qwen3.8-27B-NVFP4` | **21.9 GB** | 90.23% | 43.80% | 14.49% | 3.29% | 0.70% | 11 | **11436** |
+| `unsloth/Qwen3.8-27B-NVFP4` | 23.4 GB | 91.75% | 40.12% | 10.32% | 3.91% | 0.25% | 19 | 11069 |
 Bold marks the best value in each column **among the FP4 checkpoints**; the FP8 row is a
 reference at a different precision and size class, so it is excluded from the comparison.
+All sizes are on-disk tensor bytes and **include the ~0.85 GB BF16 MTP head**, which every
+checkpoint in this table ships. Subtract ~0.85 GB for a no-MTP comparison.
 **Columns.** `top-1` is raw argmax agreement with BF16. The four bucket columns are
 *disagreement* rates, split by how confident the base model was at that position
 ```
 `Qwen3_5ForConditionalGeneration` does not carry `mtp.*` in its state dict, so
+llm-compressor never sees it and it is silently dropped, even though `config.json` still
+declares `mtp_num_hidden_layers: 1`. It is grafted back in here from the base checkpoint
+and excluded from quantization (`re:.*mtp.*` in
 `quantization_config.ignore`; without that exclusion the quantization target regexes also
 match `mtp.layers.0.mlp.*` and vLLM fails to load). Draft quality drives acceptance rate,
 so it is kept at full precision rather than quantized.