Qwen3 235B A22B INT4 W4A16

🚀 Qwen3-235B-A22B

This repository contains the Qwen 3 235B A22B model quantized to INT4 / W4A16 by JustinJJA. It significantly reduces the model's memory footprint while maintaining accuracy.

Metadata

🚀 Quick Start

The code of Qwen3-MoE has been integrated into the latest Hugging Face transformers. We strongly recommend using the latest version of transformers. If you use transformers<4.51.0, you'll encounter the following error:

KeyError: 'qwen3_moe'

Here is a code snippet demonstrating how to use the model to generate content based on given inputs:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Qwen/Qwen3-235B-A22B"

# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

# prepare the model input
prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True # Switches between thinking and non-thinking modes. Default is True.
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# conduct text completion
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=32768
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 

# parsing thinking content
try:
    # rindex finding 151668 (</think>)
    index = len(output_ids) - output_ids[::-1].index(151668)
except ValueError:
    index = 0

thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")

print("thinking content:", thinking_content)
print("content:", content)

For deployment, you can use sglang>=0.4.6.post1 or vllm>=0.8.5 to create an OpenAI-compatible API endpoint:

• SGLang:

  python -m sglang.launch_server --model-path Qwen/Qwen3-235B-A22B --reasoning-parser qwen3
  

• vLLM:

  vllm serve Qwen/Qwen3-235B-A22B --enable-reasoning --reasoning-parser deepseek_r1
  

For local use, applications such as Ollama, LMStudio, MLX-LM, llama.cpp, and KTransformers have also supported Qwen3.

✨ Features

Qwen3 Highlights

Qwen3 is the latest generation of large language models in the Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features:

• Unique Seamless Switching: It uniquely supports seamless switching between thinking mode (for complex logical reasoning, math, and coding) and non-thinking mode (for efficient, general-purpose dialogue) within a single model, ensuring optimal performance across various scenarios.
• Enhanced Reasoning Capabilities: It significantly enhances its reasoning capabilities, surpassing previous QwQ (in thinking mode) and Qwen2.5 instruct models (in non-thinking mode) on mathematics, code generation, and commonsense logical reasoning.
• Superior Human Preference Alignment: It excels in creative writing, role-playing, multi-turn dialogues, and instruction following, delivering a more natural, engaging, and immersive conversational experience.
• Expertise in Agent Capabilities: It has expertise in agent capabilities, enabling precise integration with external tools in both thinking and non-thinking modes and achieving leading performance among open-source models in complex agent-based tasks.
• Multilingual Support: It supports 100+ languages and dialects with strong capabilities for multilingual instruction following and translation.

📦 Installation

• Inference: Use vllm>=0.8.5
  • -tp 4 max, use -tp 4 -pp 2 for 8 GPUs
  • Example command: VLLM serve justinjja/Qwen3-235B-A22B-INT4-W4A16 -tp 4
• Quantization code: Can be found here: https://huggingface.co/justinjja/Qwen3-235B-A22B-INT4-W4A16/blob/main/quantization_script.py

📚 Documentation

Model Overview

Qwen3-235B-A22B has the following features:

• Type: Causal Language Models
• Training Stage: Pretraining & Post-training
• Number of Parameters: 235B in total and 22B activated
• Number of Parameters (Non-Embedding): 234B
• Number of Layers: 94
• Number of Attention Heads (GQA): 64 for Q and 4 for KV
• Number of Experts: 128
• Number of Activated Experts: 8
• Context Length: 32,768 natively and 131,072 tokens with YaRN

For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our blog, GitHub, and Documentation.

Switching Between Thinking and Non-Thinking Mode

⚠️ Important Note

The enable_thinking switch is also available in APIs created by SGLang and vLLM. Please refer to our documentation for SGLang and vLLM users.

enable_thinking=True

By default, Qwen3 has thinking capabilities enabled, similar to QwQ-32B. This means the model will use its reasoning abilities to enhance the quality of generated responses. For example, when explicitly setting enable_thinking=True or leaving it as the default value in tokenizer.apply_chat_template, the model will engage its thinking mode.

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True  # True is the default value for enable_thinking
)

In this mode, the model will generate think content wrapped in a <think>...</think> block, followed by the final response.

⚠️ Important Note

For thinking mode, use Temperature=0.6, TopP=0.95, TopK=20, and MinP=0 (the default setting in generation_config.json). DO NOT use greedy decoding, as it can lead to performance degradation and endless repetitions. For more detailed guidance, please refer to the Best Practices section.

enable_thinking=False

We provide a hard switch to strictly disable the model's thinking behavior, aligning its functionality with the previous Qwen2.5-Instruct models. This mode is particularly useful in scenarios where disabling thinking is essential for enhancing efficiency.

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=False  # Setting enable_thinking=False disables thinking mode
)

In this mode, the model will not generate any think content and will not include a <think>...</think> block.

⚠️ Important Note

For non-thinking mode, we suggest using Temperature=0.7, TopP=0.8, TopK=20, and MinP=0. For more detailed guidance, please refer to the Best Practices section.

Advanced Usage: Switching Between Thinking and Non-Thinking Modes via User Input

We provide a soft switch mechanism that allows users to dynamically control the model's behavior when enable_thinking=True. Specifically, you can add /think and /no_think to user prompts or system messages to switch the model's thinking mode from turn to turn. The model will follow the most recent instruction in multi-turn conversations.

from transformers import AutoModelForCausalLM, AutoTokenizer

class QwenChatbot:
    def __init__(self, model_name="Qwen/Qwen3-235B-A22B"):
        self.tokenizer = AutoTokenizer.from_pretrained(model_name)
        self.model = AutoModelForCausalLM.from_pretrained(model_name)
        self.history = []

    def generate_response(self, user_input):
        messages = self.history + [{"role": "user", "content": user_input}]

        text = self.tokenizer.apply_chat_template(
            messages,
            tokenize=False,
            add_generation_prompt=True
        )

        inputs = self.tokenizer(text, return_tensors="pt")
        response_ids = self.model.generate(**inputs, max_new_tokens=32768)[0][len(inputs.input_ids[0]):].tolist()
        response = self.tokenizer.decode(response_ids, skip_special_tokens=True)

        # Update history
        self.history.append({"role": "user", "content": user_input})
        self.history.append({"role": "assistant", "content": response})

        return response

# Example Usage
if __name__ == "__main__":
    chatbot = QwenChatbot()

    # First input (without /think or /no_think tags, thinking mode is enabled by default)
    user_input_1 = "How many r's in strawberries?"
    print(f"User: {user_input_1}")
    response_1 = chatbot.generate_response(user_input_1)
    print(f"Bot: {response_1}")
    print("----------------------")

    # Second input with /no_think
    user_input_2 = "Then, how many r's in blueberries? /no_think"
    print(f"User: {user_input_2}")
    response_2 = chatbot.generate_response(user_input_2)
    print(f"Bot: {response_2}") 
    print("----------------------")

    # Third input with /think
    user_input_3 = "Really? /think"
    print(f"User: {user_input_3}")
    response_3 = chatbot.generate_response(user_input_3)
    print(f"Bot: {response_3}")

⚠️ Important Note

For API compatibility, when enable_thinking=True, regardless of whether the user uses /think or /no_think, the model will always output a block wrapped in <think>...</think>. However, the content inside this block may be empty if thinking is disabled. When enable_thinking=False, the soft switches are not valid. Regardless of any /think or /no_think tags input by the user, the model will not generate think content and will not include a <think>...</think> block.

Agentic Use

Qwen3 excels in tool calling capabilities. We recommend using Qwen-Agent to make the best use of the agentic ability of Qwen3. Qwen-Agent encapsulates tool-calling templates and tool-calling parsers internally, greatly reducing coding complexity.

from qwen_agent.agents import Assistant

# Define LLM
llm_cfg = {
    'model': 'Qwen3-235B-A22B',

    # Use the endpoint provided by Alibaba Model Studio:
    # 'model_type': 'qwen_dashscope',
    # 'api_key': os.getenv('DASHSCOPE_API_KEY'),

    # Use a custom endpoint compatible with OpenAI API:
    'model_server': 'http://localhost:8000/v1',  # api_base
    'api_key': 'EMPTY',

    # Other parameters:
    # 'generate_cfg': {
    #         # Add: When the response content is `<think>this is the thought</think>this is the answer;
    #         # Do not add: When the response has been separated by reasoning_content and content.
    #         'thought_in_content': True,
    #     },
}

# Define Tools
tools = [
    {'mcpServers': {  # You can specify the MCP configuration file
            'time': {
                'command': 'uvx',
                'args': ['mcp-server-time', '--local-timezone=Asia/Shanghai']
            },
            "fetch": {
                "command": "uvx",
                "args": ["mcp-server-fetch"]
            }
        }
    },
  'code_interpreter',  # Built-in tools
]

# Define Agent
bot = Assistant(llm=llm_cfg, function_list=tools)

# Streaming generation
messages = [{'role': 'user', 'content': 'https://qwenlm.github.io/blog/ Introduce the latest developments of Qwen'}]
for responses in bot.run(messages=messages):
    pass
print(responses)

Processing Long Texts

Qwen3 natively supports context lengths of up to 32,768 tokens. For conversations where the total length (including both input and output) significantly exceeds this limit, we recommend using RoPE scaling techniques to handle long texts effectively. We have validated the model's performance on context lengths of up to 131,072 tokens using the YaRN method.

YaRN is currently supported by several inference frameworks, e.g., transformers and llama.cpp for local use, vllm and sglang for deployment. In general, there are two approaches to enabling YaRN for supported frameworks:

• Modifying the model files: In the config.json file, add the rope_scaling fields:

  {
      ...,
      "rope_scaling": {
          "rope_type": "yarn",
          "factor": 4.0,
          "original_max_position_embeddings": 32768
      }
  }
  

📄 License

This project is licensed under the Apache-2.0 license.