README.md

---
base_model: Qwen/Qwen2.5-72B-Instruct
inference: false
language:
- en
library_name: gguf
license: other
license_link: https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/LICENSE
license_name: qwen
pipeline_tag: text-generation
quantized_by: legraphista
tags:
- chat
- quantized
- GGUF
- quantization
- imat
- imatrix
- static
- 16bit
- 8bit
- 6bit
- 5bit
- 4bit
- 3bit
- 2bit
- 1bit
---

# Qwen2.5-72B-Instruct-IMat-GGUF
_Llama.cpp imatrix quantization of Qwen/Qwen2.5-72B-Instruct_

Original Model: [Qwen/Qwen2.5-72B-Instruct](https://huggingface.co/Qwen/Qwen2.5-72B-Instruct)    
Original dtype: `BF16` (`bfloat16`)  
Quantized by:  llama.cpp [b3787](https://github.com/ggerganov/llama.cpp/releases/tag/b3787)  
IMatrix dataset: [here](https://gist.githubusercontent.com/bartowski1182/eb213dccb3571f863da82e99418f81e8/raw/b2869d80f5c16fd7082594248e80144677736635/calibration_datav3.txt)  

- [Files](#files)
    - [IMatrix](#imatrix)
    - [Common Quants](#common-quants)
    - [All Quants](#all-quants)
- [Downloading using huggingface-cli](#downloading-using-huggingface-cli)
- [Inference](#inference)
    - [Simple chat template](#simple-chat-template)
    - [Chat template with system prompt](#chat-template-with-system-prompt)
    - [Llama.cpp](#llama-cpp)
- [FAQ](#faq)
    - [Why is the IMatrix not applied everywhere?](#why-is-the-imatrix-not-applied-everywhere)
    - [How do I merge a split GGUF?](#how-do-i-merge-a-split-gguf)

---

## Files

### IMatrix
Status: ⏳ Processing  
Link: [here](https://huggingface.co/legraphista/Qwen2.5-72B-Instruct-IMat-GGUF/blob/main/imatrix.dat)

### Common Quants
| Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
| -------- | ---------- | --------- | ------ | ------------ | -------- |
| Qwen2.5-72B-Instruct.Q8_0 | Q8_0 | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q6_K | Q6_K | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q4_K | Q4_K | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q3_K | Q3_K | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q2_K | Q2_K | - | ⏳ Processing | 🟢 IMatrix | -


### All Quants
| Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
| -------- | ---------- | --------- | ------ | ------------ | -------- |
| Qwen2.5-72B-Instruct.BF16 | BF16 | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.FP16 | F16 | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q8_0 | Q8_0 | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q6_K | Q6_K | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q5_K | Q5_K | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q5_K_S | Q5_K_S | - | ⏳ Processing | ⚪ Static | -
| Qwen2.5-72B-Instruct.Q4_K | Q4_K | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q4_K_S | Q4_K_S | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ4_NL | IQ4_NL | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ4_XS | IQ4_XS | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q3_K | Q3_K | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q3_K_L | Q3_K_L | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q3_K_S | Q3_K_S | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ3_M | IQ3_M | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ3_S | IQ3_S | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ3_XS | IQ3_XS | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ3_XXS | IQ3_XXS | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q2_K | Q2_K | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.Q2_K_S | Q2_K_S | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ2_M | IQ2_M | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ2_S | IQ2_S | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ2_XS | IQ2_XS | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ2_XXS | IQ2_XXS | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ1_M | IQ1_M | - | ⏳ Processing | 🟢 IMatrix | -
| Qwen2.5-72B-Instruct.IQ1_S | IQ1_S | - | ⏳ Processing | 🟢 IMatrix | -


## 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/Qwen2.5-72B-Instruct-IMat-GGUF --include "Qwen2.5-72B-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/Qwen2.5-72B-Instruct-IMat-GGUF --include "Qwen2.5-72B-Instruct.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's
```

---

## Inference

### Simple chat template
```
<|im_start|>system
You are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|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 Qwen2.5-72B-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](https://www.reddit.com/r/LocalLLaMA/comments/1993iro/ggufs_quants_can_punch_above_their_weights_now/), 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
    - To get hold of `gguf-split`, navigate to https://github.com/ggerganov/llama.cpp/releases
    - Download the appropriate zip for your system from the latest release
    - Unzip the archive and you should be able to find `gguf-split`
2. Locate your GGUF chunks folder (ex: `Qwen2.5-72B-Instruct.Q8_0`)
3. Run `gguf-split --merge Qwen2.5-72B-Instruct.Q8_0/Qwen2.5-72B-Instruct.Q8_0-00001-of-XXXXX.gguf Qwen2.5-72B-Instruct.Q8_0.gguf`
    - Make sure to point `gguf-split` to the first chunk of the split.

---

Got a suggestion? Ping me [@legraphista](https://x.com/legraphista)!