Microsoft Phi 4 Reasoning Plus GGUF

🚀 Llamacpp imatrix Quantizations of Phi-4-reasoning-plus by microsoft

This project provides Llama.cpp imatrix quantizations of the Phi-4-reasoning-plus model by Microsoft. It uses specific releases of Llama.cpp for quantization and offers various quantized models with different characteristics.

✨ Features

• Quantization: Utilizes llama.cpp release b5228 for quantization.
• Multiple Quantized Models: Offers a wide range of quantized models with different file sizes, qualities, and performance characteristics.
• Prompt Format: Defines a specific prompt format for interacting with the model.
• Download Options: Provides instructions for downloading models using the Hugging Face CLI.
• ARM/AVX Support: Includes information on ARM/AVX performance and online repacking of weights.

📦 Installation

Prerequisites

Make sure you have the Hugging Face CLI installed:

pip install -U "huggingface_hub[cli]"

Downloading Models

• Single File: To download a specific file, run the following command:

huggingface-cli download bartowski/microsoft_Phi-4-reasoning-plus-GGUF --include "microsoft_Phi-4-reasoning-plus-Q4_K_M.gguf" --local-dir ./

• Multiple Files: If the model is split into multiple files (bigger than 50GB), run:

huggingface-cli download bartowski/microsoft_Phi-4-reasoning-plus-GGUF --include "microsoft_Phi-4-reasoning-plus-Q8_0/*" --local-dir ./

💻 Usage Examples

Prompt Format

<|im_start|>system<|im_sep|>You are Phi, a language model trained by Microsoft to help users. Your role as an assistant involves thoroughly exploring questions through a systematic thinking process before providing the final precise and accurate solutions. This requires engaging in a comprehensive cycle of analysis, summarizing, exploration, reassessment, reflection, backtracing, and iteration to develop well-considered thinking process. Please structure your response into two main sections: Thought and Solution using the specified format:<think>{Thought section}</think>{Solution section}. In the Thought section, detail your reasoning process in steps. Each step should include detailed considerations such as analysing questions, summarizing relevant findings, brainstorming new ideas, verifying the accuracy of the current steps, refining any errors, and revisiting previous steps. In the Solution section, based on various attempts, explorations, and reflections from the Thought section, systematically present the final solution that you deem correct. The Solution section should be logical, accurate, and concise and detail necessary steps needed to reach the conclusion. Now, try to solve the following question through the above guidelines:<|im_end|>{system_prompt}<|end|><|user|>{prompt}<|end|><|assistant|>

📚 Documentation

Model Information

• Original Model: https://huggingface.co/microsoft/Phi-4-reasoning-plus
• Quantization Method: All quants are made using the imatrix option with a dataset from here.

Model Download Options

Embed/Output Weights

Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.

ARM/AVX Information

Previously, you would download Q4_0_4_4/4_8/8_8, and these would have their weights interleaved in memory in order to improve performance on ARM and AVX machines by loading up more data in one pass.

Now, however, there is something called "online repacking" for weights. details in this PR. If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly.

As of llama.cpp build b4282 you will not be able to run the Q4_0_X_X files and will instead need to use Q4_0.

Additionally, if you want to get slightly better quality for , you can use IQ4_NL thanks to this PR which will also repack the weights for ARM, though only the 4_4 for now. The loading time may be slower but it will result in an overall speed incrase.

Which File to Choose

🔧 Technical Details

Quantization Process

The quantization is performed using llama.cpp release b5228. All quants are made using the imatrix option with a dataset from here.

Online Repacking

The concept of "online repacking" for weights is introduced in this PR. If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly.

📄 License

This project is licensed under the MIT License. You can find the license details here.

Credits

• Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.
• Thank you ZeroWw for the inspiration to experiment with embed/output.
• Thank you to LM Studio for sponsoring my work.

If you want to support my work, visit my ko-fi page here: https://ko-fi.com/bartowski