Instructions to use YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound with Transformers:

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

pipe = pipeline("text-generation", model="YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound")
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("YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound")
model = AutoModelForCausalLM.from_pretrained("YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound", 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 YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound

SGLang

How to use YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound 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 "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound" \
    --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": "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound",
		"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 "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound" \
        --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": "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound with Docker Model Runner:
```
docker model run hf.co/YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound
```

YCWTG commited on Feb 25

Commit

fc3d512

verified ·

1 Parent(s): 11bca80

Update README.md

Browse files

Files changed (1) hide show

README.md +8 -11

README.md CHANGED Viewed

@@ -18,7 +18,7 @@ pipeline_tag: text-generation
 语言 [中文](https://huggingface.co/YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound/blob/main/README_zh.md)|English
 ## Model Details
-This model is an int2 model with group_size 512 and symmetric quantization of [Qwen/Qwen3-Coder-Next](https://huggingface.co/Qwen/Qwen3-Coder-Next) generated by [intel/auto-round](https://github.com/intel/auto-round). Please follow the license of the original model.
 ### Quantization Strategy (Intel MoE Recipe)
@@ -29,21 +29,18 @@ This model is an int2 model with group_size 512 and symmetric quantization of [Q
 | shared_expert_gate | 16-bit | Skipped (shape not divisible by 32) |
 | lm_head | Original | Excluded by AutoRound |
-### MMLU-Pro
-| Model | Accuracy | Delta |
-|-------|----------|-------|
-| BF16 | 52.90% | - |
-| **W2A16** | **51.27%** | **-1.63%** |
-## How to Use
 ### HF Usage
 ```python
 from transformers import AutoModelForCausalLM, AutoTokenizer
-model_name = "YCWTG/Qwen3-Coder-Next-int2-AutoRound-best"
 # load the tokenizer and the model
 tokenizer = AutoTokenizer.from_pretrained(model_name)
@@ -75,7 +72,7 @@ print("content:", content)
 ```bash
 from auto_round import AutoRound
-model_name = "/home/ycwtg/PycharmProjects/MMLU-Pro/~/.cache/model/Qwen3-Coder-Next"
 # Build layer config for mixed-bits (Intel recipe)
 layer_config = {}
@@ -108,7 +105,7 @@ autoround = AutoRound(
     low_gpu_mem_usage=True,
     enable_alg_ext=True
 )
-output_dir="/home/ycwtg/PycharmProjects/MMLU-Pro/~/.cache/model/Qwen3-Coder-Next-int2-mixed-AutoRound"
 autoround.quantize_and_save(output_dir,format="auto_round" )
 ```

 语言 [中文](https://huggingface.co/YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound/blob/main/README_zh.md)|English
 ## Model Details
+This model is an **mixed-bits INT2 quantized** model with group_size 512 and symmetric quantization of [Qwen/Qwen3-Coder-Next](https://huggingface.co/Qwen/Qwen3-Coder-Next) generated by [intel/auto-round](https://github.com/intel/auto-round). Please follow the license of the original model.
 ### Quantization Strategy (Intel MoE Recipe)
 | shared_expert_gate | 16-bit | Skipped (shape not divisible by 32) |
 | lm_head | Original | Excluded by AutoRound |
+### Model Size
+- **Original BF16**: ~160GB
+- **mixed INT2**: ~25GB
+## Quickstart
 ### HF Usage
 ```python
 from transformers import AutoModelForCausalLM, AutoTokenizer
+model_name = "YCWTG/Qwen3-Coder-Next-int2-mixed-AutoRound"
 # load the tokenizer and the model
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 ```bash
 from auto_round import AutoRound
+model_name = "Qwen/Qwen3-Coder-Next"
 # Build layer config for mixed-bits (Intel recipe)
 layer_config = {}
     low_gpu_mem_usage=True,
     enable_alg_ext=True
 )
+output_dir="~/.cache/model/Qwen3-Coder-Next-int2-mixed-AutoRound"
 autoround.quantize_and_save(output_dir,format="auto_round" )
 ```