This project provides the GGUF format file of the Qwen3-4B-Thinking-2507 model. It is a large language model with 4B parameters and chain of thought reasoning ability, supporting multiple quantization versions for easy deployment and operation on various hardware.

Natural Language ProcessingGguf

This model is a knowledge distillation model based on the Qwen3-4B architecture, trained with approximately 54.4 million tokens generated by Gemini 2.5 Flash, aiming to integrate the behavior, inference process, and knowledge of Gemini-2.5 Flash into a single dataset.

Natural Language ProcessingTransformersEnglish

This is a quantized version of the Qwen3-4B-Instruct-2507-heretic model. Multiple quantization types of model files are generated using the llama.cpp tool and the imatrix dataset, which can run in various environments and provide users with more options.

Natural Language ProcessingGguf

This is a mixed quantization version based on the Qwen3 4B Instruct model, using MXFP4 and mixed weight techniques to achieve a smaller file size and higher inference speed while maintaining nearly lossless precision.

Natural Language ProcessingSafetensorsOther

This is a mixed quantization version based on the Qwen3 4B model, using the MXFP4_MOE mixed weight technology. It achieves a smaller file size and higher inference speed while maintaining near-lossless accuracy. By carefully combining MXFP4 and high-precision embedding/output weights, the model reaches an accuracy level close to Q8 quantization and has a throughput at the Q4 - Q6 level.

Natural Language ProcessingSafetensorsOther

This model is based on Qwen3-4B-Thinking-2507 and fine-tuned on 1000 examples of GPT-5-Codex. It focuses on text generation tasks and uses Unsloth technology to achieve a 2-fold increase in training speed.

Natural Language ProcessingTransformersEnglish

This is a quantized version fine-tuned based on the Qwen3-4B model. It uses the Unsloth framework and the Huggingface TRL library for efficient training, with a 2x increase in training speed. The model adopts an int8-int4 mixed quantization scheme and supports running on mobile devices.

Natural Language ProcessingTransformersEnglish

Qwen3-4B-SafeRL is a safety-aligned version of the Qwen3-4B model. It enhances the model's robustness against harmful or adversarial prompts through reinforcement learning training. This version uses a hybrid reward function for optimization, balancing three objectives: safety, usefulness, and minimizing unnecessary rejections.

Natural Language ProcessingTransformersEnglish

Qwen3-4B-SafeRL is a safety-aligned version based on the Qwen3-4B model. It is trained through reinforcement learning and combined with the reward signals of Qwen3Guard-Gen, enhancing the model's robustness against harmful or adversarial prompts. While ensuring safety, it avoids overly simple or evasive rejection behaviors.

Natural Language ProcessingTransformers

OmniDimen-V1.1-4B-Emotion is an emotion recognition and emotion-aware text generation model fine-tuned based on Qwen3-4B-Instruct-2507. It is specifically designed to accurately identify the emotional tendency of text and generate matching emotional content, providing users with a more emotionally resonant interaction experience.

Natural Language ProcessingGgufMultiple Languages

This project is a text generation model obtained by quantizing the Qwen3-4B model using AMD's quantization tool. It can be used in scenarios such as chatting. The model is prepared using the AMD Quark quantization tool and undergoes necessary post-processing.

Natural Language ProcessingOnnxEnglish

This is a GGUF format version converted from the Qwen/Qwen3-4B-Thinking-2507 model, using Q8_0 quantization, which can be used compatibly with llama.cpp. This model has chain-of-thought reasoning ability and has 4 billion parameters.

Natural Language ProcessingTransformers

Qwen3-4B-I-1509 is a dedicated model fine-tuned based on the Qwen3-4B-Instruct-2507 base model, focusing on tool usage and function call generation tasks. This model is trained using a reinforcement learning method with multi-signal rewards and demonstrates good performance in multiple evaluation benchmarks.

Natural Language ProcessingTransformersEnglish

This is the GGUF quantized version of the Qwen/Qwen3-4B language model, a large language model with 4 billion parameters developed by Alibaba's Tongyi series. It is designed for consumer-grade hardware and supports powerful reasoning, intelligent agent workflows, and multilingual interaction.

Natural Language ProcessingGgufMultiple Languages

A professional data science agent model fine-tuned based on Qwen3-4B-Thinking-2507, specifically optimized for the Jupyter Notebook environment, capable of executing Python code, analyzing datasets, and providing step-by-step reasoning.

Natural Language ProcessingTransformersMultiple Languages

OmniDimen-4B-Emotion is an emotion recognition and emotion - aware text generation model fine - tuned based on Qwen3-4B-Instruct-2507, specifically designed for emotional interaction and dialogue generation with emotion perception ability.

Natural Language ProcessingGgufMultiple Languages

Qwen3-4B-Instruct is a large language model with 4 billion parameters developed by the Tongyi Qianwen team at Alibaba. It is optimized based on the Qwen3 architecture and specifically trained for dialogue and instruction-following tasks, with strong text generation and understanding capabilities.

Natural Language ProcessingTransformers

Qwen3 4B Instruct is a 4-billion parameter instruction fine-tuning model based on the Qwen3 architecture, supporting multilingual text generation tasks. The model offers various quantization versions, ranging from 2-bit to 32-bit, providing users with flexible choices for balancing performance and efficiency.

Natural Language ProcessingGgufMultiple Languages

Jan-v1-AIO-GGUF contains two powerful language models: Jan-v1-4B and Jan-v1-edge. Jan-v1-4B is based on the Qwen3-4B-thinking architecture, with 4 billion parameters. It has strong intelligent reasoning, problem-solving, and tool-using capabilities and supports web search tasks. Jan-v1-edge is a lightweight intelligent model designed for fast execution on devices and is suitable for resource-constrained environments.

Natural Language ProcessingTransformersEnglish

This is an MLX version memory agent model with 8-bit precision, trained based on Qwen3-4B-Thinking-2507, specifically designed to handle information retrieval, update, and clarification tasks in the memory system.

Natural Language ProcessingTransformers