axyzdong

/

AMchat-GGUF

like 0

AMchat GGUF Model

AM (Advanced Mathematics) Chat is a large-scale language model that integrates mathematical knowledge, advanced mathematics problems, and their solutions. This model utilizes a dataset that combines Math and advanced mathematics problems with their analyses. It is based on the InternLM2-Math-7B model and has been fine-tuned with xtuner, specifically designed to solve advanced mathematics problems.

Latest Release

2024-08-16

• Q6_K
• Q5_K_M
• Q5_0
• Q4_0
• Q3_K_M
• Q2_K

2024-08-09

• F16 Quantization: Achieves a balanced trade-off between model size and performance. Ideal for applications requiring precision with reduced resource consumption.
• Q8_0 Quantization: Offers a substantial reduction in model size while maintaining high accuracy, making it suitable for environments with stringent memory constraints.
• Q4_K_M Quantization: Provides the most compact model size with minimal impact on performance, perfect for deployment in resource-constrained settings.

Getting Started - Ollama

To get started with AMchat in Ollama, follow these steps:

1. Clone the Repository

   git lfs install
   git clone https://huggingface.co/axyzdong/AMchat-GGUF
   

2. Creat Model

   Make sure you have installed ollama in advance. https://ollama.com/

   ollama create AMchat -f Modelfile
   

3. Run

   ollama run AMchat
   

Getting Started - llama-cli

You can use llama-cli for conducting inference. For a detailed explanation of llama-cli, please refer to this guide

Installation

We recommend building llama.cpp from source. The following code snippet provides an example for the Linux CUDA platform. For instructions on other platforms, please refer to the official guide.

• Step 1: create a conda environment and install cmake

conda create --name AMchat python=3.10 -y
conda activate AMchat
pip install cmake

• Step 2: clone the source code and build the project

git clone --depth=1 https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build -DGGML_CUDA=ON
cmake --build build --config Release -j

All the built targets can be found in the sub directory build/bin

In the following sections, we assume that the working directory is at the root directory of llama.cpp.

Download models

You can download the appropriate model based on your requirements. For instance, AMchat-q8_0.gguf can be downloaded as below：

pip install huggingface-hub
huggingface-cli download axyzdong/AMchat-GGUF AMchat-q8_0.gguf --local-dir . --local-dir-use-symlinks False

chat example

build/bin/llama-cli \
    --model AMchat-fp16.gguf  \
    --predict 512 \
    --ctx-size 4096 \
    --gpu-layers 24 \
    --temp 0.8 \
    --top-p 0.8 \
    --top-k 50 \
    --seed 1024 \
    --color \
    --prompt "<|im_start|>system\nYou are an expert in advanced math and you can answer all kinds of advanced math problems.<|im_end|>\n" \
    --interactive \
    --multiline-input \
    --conversation \
    --verbose \
    --logdir workdir/logdir \
    --in-prefix "<|im_start|>user\n" \
    --in-suffix "<|im_end|>\n<|im_start|>assistant\n"

Star Us

If you find AMchat useful, please ⭐ Star this repository and help others discover it!