--- title: Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 canonical_url: "https://www.modelscope.cn/models/JunHowie/Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4" md_url: "https://www.modelscope.cn/models/JunHowie/Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4.md" repository: JunHowie/Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 chinese_name: Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 last_updated: 2025-12-07 license: apache-2.0 pipeline_tag: text-generation tasks: - text-generation model_type: - qwen3 architectures: - Qwen3ForCausalLM base_model: - Qwen/Qwen3-32B base_model_relation: quantized parameters: 4.6B tensor_type: - BF16 - I32 - F16 - F32 library_name: - safetensors inference_backends: - "deploy_task text/emb" - "lmdeploy_turbomind 0.9.1" - "sglang 0.5.2" - "vllm 0.9.2" downloads: 270 stars: 1 --- # Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 > Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 - JunHowie 在 ModelScope 开源的模型。Qwen3-32B-Instruct-2512-DFT DFT stands for Dynamic Fine-Tuning. For details, see this article JunHowie/Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 是 ModelScope 魔搭社区上的 4.6B 参数text-generation模型，采用 apache-2.0 许可，基于 Qwen/Qwen3-32B 构建，可用 deploy_task text/emb、lmdeploy_turbomind 0.9.1、sglang 0.5.2 部署。 - **Repository**: JunHowie/Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 - **License**: apache-2.0 - **Tasks**: text-generation - **Parameters**: 4.6B - **Base model**: Qwen/Qwen3-32B - **Inference backends**: deploy_task text/emb, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2 - **Downloads**: 270 - **Stars**: 1 - **Last updated**: 2025-12-07 Source: https://www.modelscope.cn/models/JunHowie/Qwen3-32B-Instruct-2512-DFT-GPTQ-Int4 --- # Qwen3-32B-Instruct-2512-DFT DFT stands for Dynamic Fine-Tuning. For details, [see this article](https://zhuanlan.zhihu.com/p/1940793068631589732) **NOTE：This model is the Instruct-aligned variant, and it will not generate ```` blocks in its outputs. Additionally, there is no need to specify enable_thinking=False anymore.** Among them, the 8B and 14B SFT and DFT variants are obtained via full-parameter fine-tuning, while the 32B models are trained using LoRA due to hardware resource constraints.
The dataset used is the Chinese Distillation Dataset based on Qwen3-235B-2507.available at:[**Chinese-Qwen3-235B-2507-Distill-data-110k**](https://www.modelscope.cn/datasets/swift/Chinese-Qwen3-235B-2507-Distill-data-110k)
For details and code regarding model training and quantization, please see[Training and Quantization Guide](https://www.modelscope.cn/learn/3000)
Here is the list of models released in this version:

Model	4-bit AWQ		8-bit FP8	GPTQ		NVIDIA FP4		Weight-Activation
Model	AWQ	AWQ-asym	8-bit FP8	INT4	INT8	NVFP4	NVFP4-A16	W4A16	W8A8
Qwen3-8B-Instruct-2512-DFT	AWQ	awq-asym	FP8	GPTQ(int4)	GPTQ(int8)	NVFP4	NVFP4A16	W4A16	W8A8
Qwen3-8B-Instruct-2512-SFT	AWQ	awq-asym	FP8	GPTQ(int4)	GPTQ(int8)	NVFP4	NVFP4A16	W4A16	W8A8
Qwen3-14B-Instruct-2512-DFT	AWQ	awq-asym	FP8	GPTQ(int4)	GPTQ(int8)	NVFP4	NVFP4A16	W4A16	W8A8
Qwen3-14B-Instruct-2512-SFT	AWQ	awq-asym	FP8	GPTQ(int4)	GPTQ(int8)	NVFP4	NVFP4A16	W4A16	W8A8
Qwen3-32B-Instruct-2512-DFT	AWQ	awq-asym	FP8	GPTQ(int4)	GPTQ(int8)	NVFP4	NVFP4A16	W4A16	W8A8

### 【Dependencies】 ``` vllm>=0.10.2 transformers>=4.56.1 ```