legraphista's picture
Upload README.md with huggingface_hub
86759eb verified
metadata
base_model: Qwen/Qwen2.5-Math-7B-Instruct
inference: false
language:
  - en
library_name: gguf
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen2.5-Math-7B-Instruct/blob/main/LICENSE
pipeline_tag: text-generation
quantized_by: legraphista
tags:
  - chat
  - quantized
  - GGUF
  - quantization
  - imat
  - imatrix
  - static
  - 16bit
  - 8bit
  - 6bit
  - 5bit
  - 4bit
  - 3bit
  - 2bit
  - 1bit

Qwen2.5-Math-7B-Instruct-IMat-GGUF

Llama.cpp imatrix quantization of Qwen/Qwen2.5-Math-7B-Instruct

Original Model: Qwen/Qwen2.5-Math-7B-Instruct
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3787
IMatrix dataset: here


Files

IMatrix

Status: βœ… Available
Link: here

Common Quants

Filename Quant type File Size Status Uses IMatrix Is Split
Qwen2.5-Math-7B-Instruct.Q8_0.gguf Q8_0 8.10GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q6_K.gguf Q6_K 6.25GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q4_K.gguf Q4_K 4.68GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q3_K.gguf Q3_K 3.81GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q2_K.gguf Q2_K 3.02GB βœ… Available 🟒 IMatrix πŸ“¦ No

All Quants

Filename Quant type File Size Status Uses IMatrix Is Split
Qwen2.5-Math-7B-Instruct.BF16.gguf BF16 15.24GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.FP16.gguf F16 15.24GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q8_0.gguf Q8_0 8.10GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q6_K.gguf Q6_K 6.25GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q5_K.gguf Q5_K 5.44GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q5_K_S.gguf Q5_K_S 5.32GB βœ… Available βšͺ Static πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q4_K.gguf Q4_K 4.68GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q4_K_S.gguf Q4_K_S 4.46GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ4_NL.gguf IQ4_NL 4.44GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ4_XS.gguf IQ4_XS 4.22GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q3_K.gguf Q3_K 3.81GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q3_K_L.gguf Q3_K_L 4.09GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q3_K_S.gguf Q3_K_S 3.49GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ3_M.gguf IQ3_M 3.57GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ3_S.gguf IQ3_S 3.50GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ3_XS.gguf IQ3_XS 3.35GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ3_XXS.gguf IQ3_XXS 3.11GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q2_K.gguf Q2_K 3.02GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.Q2_K_S.gguf Q2_K_S 2.83GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ2_M.gguf IQ2_M 2.78GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ2_S.gguf IQ2_S 2.60GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ2_XS.gguf IQ2_XS 2.47GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ2_XXS.gguf IQ2_XXS 2.27GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ1_M.gguf IQ1_M 2.04GB βœ… Available 🟒 IMatrix πŸ“¦ No
Qwen2.5-Math-7B-Instruct.IQ1_S.gguf IQ1_S 1.90GB βœ… Available 🟒 IMatrix πŸ“¦ No

Downloading using huggingface-cli

If you do not have hugginface-cli installed:

pip install -U "huggingface_hub[cli]"

Download the specific file you want:

huggingface-cli download legraphista/Qwen2.5-Math-7B-Instruct-IMat-GGUF --include "Qwen2.5-Math-7B-Instruct.Q8_0.gguf" --local-dir ./

If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download legraphista/Qwen2.5-Math-7B-Instruct-IMat-GGUF --include "Qwen2.5-Math-7B-Instruct.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

Inference

Simple chat template

<|im_start|>system
Please reason step by step, and put your final answer within \boxed{}.<|im_end|>
<|im_start|>user
{user_prompt}<|im_end|>
<|im_start|>assistant
{assistant_response}<|im_end|>
<|im_start|>user
{next_user_prompt}<|im_end|>

Chat template with system prompt

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{user_prompt}<|im_end|>
<|im_start|>assistant
{assistant_response}<|im_end|>
<|im_start|>user
{next_user_prompt}<|im_end|>

Llama.cpp

llama.cpp/main -m Qwen2.5-Math-7B-Instruct.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"

FAQ

Why is the IMatrix not applied everywhere?

According to this investigation, it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).

How do I merge a split GGUF?

  1. Make sure you have gguf-split available
  2. Locate your GGUF chunks folder (ex: Qwen2.5-Math-7B-Instruct.Q8_0)
  3. Run gguf-split --merge Qwen2.5-Math-7B-Instruct.Q8_0/Qwen2.5-Math-7B-Instruct.Q8_0-00001-of-XXXXX.gguf Qwen2.5-Math-7B-Instruct.Q8_0.gguf
    • Make sure to point gguf-split to the first chunk of the split.

Got a suggestion? Ping me @legraphista!