Qwen2-7B-Instruct-IMat-GGUF

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

Original Model: Qwen/Qwen2-7B-Instruct
Original dtype: BF16 (bfloat16)
Quantized by: llama.cpp b3091
IMatrix dataset: here

• Files
  • IMatrix
  • Common Quants
  • All Quants
• Downloading using huggingface-cli
• Inference
  • Simple chat template
  • Chat template with system prompt
  • Llama.cpp
• FAQ
  • Why is the IMatrix not applied everywhere?
  • How do I merge a split GGUF?

---

Files

IMatrix

Status: ✅ Available
Link: here

Common Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
Qwen2-7B-Instruct.Q8_0.gguf	Q8_0	8.10GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q6_K.gguf	Q6_K	6.25GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q4_K.gguf	Q4_K	4.68GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q3_K.gguf	Q3_K	3.81GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q2_K.gguf	Q2_K	3.02GB	✅ Available	🟢 IMatrix	📦 No

All Quants

Filename	Quant type	File Size	Status	Uses IMatrix	Is Split
Qwen2-7B-Instruct.BF16.gguf	BF16	15.24GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.FP16.gguf	F16	15.24GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q8_0.gguf	Q8_0	8.10GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q6_K.gguf	Q6_K	6.25GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q5_K.gguf	Q5_K	5.44GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q5_K_S.gguf	Q5_K_S	5.32GB	✅ Available	⚪ Static	📦 No
Qwen2-7B-Instruct.Q4_K.gguf	Q4_K	4.68GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q4_K_S.gguf	Q4_K_S	4.46GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ4_NL.gguf	IQ4_NL	4.44GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ4_XS.gguf	IQ4_XS	4.22GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q3_K.gguf	Q3_K	3.81GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q3_K_L.gguf	Q3_K_L	4.09GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q3_K_S.gguf	Q3_K_S	3.49GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ3_M.gguf	IQ3_M	3.57GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ3_S.gguf	IQ3_S	3.50GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ3_XS.gguf	IQ3_XS	3.35GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ3_XXS.gguf	IQ3_XXS	3.11GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q2_K.gguf	Q2_K	3.02GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.Q2_K_S.gguf	Q2_K_S	2.83GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ2_M.gguf	IQ2_M	2.78GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ2_S.gguf	IQ2_S	2.60GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ2_XS.gguf	IQ2_XS	2.47GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ2_XXS.gguf	IQ2_XXS	2.27GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ1_M.gguf	IQ1_M	2.04GB	✅ Available	🟢 IMatrix	📦 No
Qwen2-7B-Instruct.IQ1_S.gguf	IQ1_S	1.90GB	✅ Available	🟢 IMatrix	📦 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/Qwen2-7B-Instruct-IMat-GGUF --include "Qwen2-7B-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/Qwen2-7B-Instruct-IMat-GGUF --include "Qwen2-7B-Instruct.Q8_0/*" --local-dir ./
# see FAQ for merging GGUF's

---

Inference

Simple chat template

<|im_start|>system
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 Qwen2-7B-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
   • 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: Qwen2-7B-Instruct.Q8_0)
3. Run gguf-split --merge Qwen2-7B-Instruct.Q8_0/Qwen2-7B-Instruct.Q8_0-00001-of-XXXXX.gguf Qwen2-7B-Instruct.Q8_0.gguf
   • Make sure to point gguf-split to the first chunk of the split.

---

Got a suggestion? Ping me @legraphista!