mlabonne_Qwen3-1.7B-abliterated-GGUF Open Source Model - A Text Generation Tool Supporting Multiple Types of Quantization

Mlabonne Qwen3 1.7B Abliterated GGUF

Developed by bartowski

This is a quantized version based on the Qwen3-1.7B-abliterated model, using llama.cpp for quantization, supporting multiple quantization types, suitable for text generation tasks.

Large Language Model #Multi-quantization versions #Lightweight generation #Chinese dialogue optimization

Downloads 1,493

Release Time : 4/30/2025

Model Overview

This model is a quantized version of Qwen3-1.7B-abliterated, suitable for text generation tasks, supporting multiple quantization types, and can be run in LM Studio or llama.cpp.

Model Features

Multiple quantization options

Offers various quantization versions from BF16 to Q2_K to meet different hardware needs.

imatrix quantization

Uses imatrix option for quantization to enhance model performance.

Supports multiple runtime environments

Can be run in LM Studio, llama.cpp, or any project based on llama.cpp.

Embedding/output weight optimization

In some quantized versions, embedding and output weights are quantized to Q8_0 for improved performance.

Model Capabilities

Text generation

Multilingual support

Quantized inference

Use Cases

Text generation

Dialogue systems

Can be used to build dialogue systems for generating natural language responses.

Content creation

Assists in generating articles, stories, and other content.

🚀 Llamacpp imatrix Quantizations of Qwen3-1.7B-abliterated by mlabonne

This project provides llama.cpp imatrix quantizations of the Qwen3-1.7B-abliterated model. It uses specific llama.cpp releases for quantization and offers various quantized models for different usage scenarios.

🚀 Quick Start

Running the Model

Using LM Studio: You can run the quantized models in LM Studio.
Using llama.cpp: Run the models directly with llama.cpp, or any other llama.cpp based project.

Prompt Format

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Downloading the Model

You can download a specific file from the following table. Here are the steps to download using huggingface-cli:

First, make sure you have huggingface-cli installed:

pip install -U "huggingface_hub[cli]"

Then, you can target the specific file you want:

huggingface-cli download bartowski/mlabonne_Qwen3-1.7B-abliterated-GGUF --include "mlabonne_Qwen3-1.7B-abliterated-Q4_K_M.gguf" --local-dir ./

If the model is bigger than 50GB, it will have been split into multiple files. In order to download them all to a local folder, run:

huggingface-cli download bartowski/mlabonne_Qwen3-1.7B-abliterated-GGUF --include "mlabonne_Qwen3-1.7B-abliterated-Q8_0/*" --local-dir ./

You can either specify a new local-dir (e.g., mlabonne_Qwen3-1.7B-abliterated-Q8_0) or download them all in place (e.g., ./).

✨ Features

Quantization Details

Quantization Tool: Using llama.cpp release b5228 for quantization.
Original Model: The original model can be found at https://huggingface.co/mlabonne/Qwen3-1.7B-abliterated.
Quantization Option: All quants are made using the imatrix option with a dataset from here.

Available Quantized Models

Filename	Quant type	File Size	Split	Description
Qwen3-1.7B-abliterated-bf16.gguf	bf16	3.45GB	false	Full BF16 weights.
Qwen3-1.7B-abliterated-Q8_0.gguf	Q8_0	1.83GB	false	Extremely high quality, generally unneeded but max available quant.
Qwen3-1.7B-abliterated-Q6_K_L.gguf	Q6_K_L	1.49GB	false	Uses Q8_0 for embed and output weights. Very high quality, near perfect, recommended.
Qwen3-1.7B-abliterated-Q6_K.gguf	Q6_K	1.42GB	false	Very high quality, near perfect, recommended.
Qwen3-1.7B-abliterated-Q5_K_L.gguf	Q5_K_L	1.33GB	false	Uses Q8_0 for embed and output weights. High quality, recommended.
Qwen3-1.7B-abliterated-Q5_K_M.gguf	Q5_K_M	1.26GB	false	High quality, recommended.
Qwen3-1.7B-abliterated-Q5_K_S.gguf	Q5_K_S	1.23GB	false	High quality, recommended.
Qwen3-1.7B-abliterated-Q4_K_L.gguf	Q4_K_L	1.18GB	false	Uses Q8_0 for embed and output weights. Good quality, recommended.
Qwen3-1.7B-abliterated-Q4_1.gguf	Q4_1	1.14GB	false	Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon.
Qwen3-1.7B-abliterated-Q4_K_M.gguf	Q4_K_M	1.11GB	false	Good quality, default size for most use cases, recommended.
Qwen3-1.7B-abliterated-Q3_K_XL.gguf	Q3_K_XL	1.08GB	false	Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability.
Qwen3-1.7B-abliterated-Q4_K_S.gguf	Q4_K_S	1.06GB	false	Slightly lower quality with more space savings, recommended.
Qwen3-1.7B-abliterated-Q4_0.gguf	Q4_0	1.06GB	false	Legacy format, offers online repacking for ARM and AVX CPU inference.
Qwen3-1.7B-abliterated-IQ4_NL.gguf	IQ4_NL	1.05GB	false	Similar to IQ4_XS, but slightly larger. Offers online repacking for ARM CPU inference.
Qwen3-1.7B-abliterated-IQ4_XS.gguf	IQ4_XS	1.01GB	false	Decent quality, smaller than Q4_K_S with similar performance, recommended.
Qwen3-1.7B-abliterated-Q3_K_L.gguf	Q3_K_L	1.00GB	false	Lower quality but usable, good for low RAM availability.
Qwen3-1.7B-abliterated-Q3_K_M.gguf	Q3_K_M	0.94GB	false	Low quality.
Qwen3-1.7B-abliterated-IQ3_M.gguf	IQ3_M	0.90GB	false	Medium-low quality, new method with decent performance comparable to Q3_K_M.
Qwen3-1.7B-abliterated-Q3_K_S.gguf	Q3_K_S	0.87GB	false	Low quality, not recommended.
Qwen3-1.7B-abliterated-Q2_K_L.gguf	Q2_K_L	0.85GB	false	Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable.
Qwen3-1.7B-abliterated-IQ3_XS.gguf	IQ3_XS	0.83GB	false	Lower quality, new method with decent performance, slightly better than Q3_K_S.
Qwen3-1.7B-abliterated-Q2_K.gguf	Q2_K	0.78GB	false	Very low quality but surprisingly usable.
Qwen3-1.7B-abliterated-IQ3_XXS.gguf	IQ3_XXS	0.75GB	false	Lower quality, new method with decent performance, comparable to Q3 quants.
Qwen3-1.7B-abliterated-IQ2_M.gguf	IQ2_M	0.70GB	false	Relatively low quality, uses SOTA techniques to be surprisingly usable.

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 can be found 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, 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 increase.

Click to view Q4_0_X_X information (deprecated)

I'm keeping this section to show the potential theoretical uplift in performance from using the Q4_0 with online repacking.

Click to view benchmarks on an AVX2 system (EPYC7702)

| model                          |       size |     params | backend    | threads |          test |                  t/s |  % (vs Q4_0)  |
| ------------------------------ | ---------: | ---------: | ---------- | ------: | ------------: | -------------------: |-------------: |
| qwen2 3B Q4_0                  |   1.70 GiB |     3.09 B | CPU        |      64 |         pp512 |        204.03 ± 1.03 |          100% |
| qwen2 3B Q4_0                  |   1.70 GiB |     3.09 B | CPU        |      64 |        pp1024 |        282.92 ± 0.19 |          100% |
| qwen2 3B Q4_0                  |   1.70 GiB |     3.09 B | CPU        |      64 |        pp2048 |        259.49 ± 0.44 |          100% |
| qwen2 3B Q4_0                  |   1.70 GiB |     3.09 B | CPU        |      64 |         tg128 |         39.12 ± 0.27 |          100% |
| qwen2 3B Q4_0                  |   1.70 GiB |     3.09 B | CPU        |      64 |         tg256 |         39.31 ± 0.69 |          100% |
| qwen2 3B Q4_0                  |   1.70 GiB |     3.09 B | CPU        |      64 |         tg512 |         40.52 ± 0.03 |          100% |
| qwen2 3B Q4_K_M                |   1.79 GiB |     3.09 B | CPU        |      64 |         pp512 |        301.02 ± 1.74 |          147% |
| qwen2 3B Q4_K_M                |   1.79 GiB |     3.09 B | CPU        |      64 |        pp1024 |        287.23 ± 0.20 |          101% |
| qwen2 3B Q4_K_M                |   1.79 GiB |     3.09 B | CPU        |      64 |        pp2048 |        262.77 ± 1.81 |          101% |
| qwen2 3B Q4_K_M                |   1.79 GiB |     3.09 B | CPU        |      64 |         tg128 |         18.80 ± 0.99 |           48% |
| qwen2 3B Q4_K_M                |   1.79 GiB |     3.09 B | CPU        |      64 |         tg256 |         24.46 ± 3.04 |           83% |
| qwen2 3B Q4_K_M                |   1.79 GiB |     3.09 B | CPU        |      64 |         tg512 |         36.32 ± 3.59 |           90% |
| qwen2 3B Q4_0_8_8              |   1.69 GiB |     3.09 B | CPU        |      64 |         pp512 |        271.71 ± 3.53 |          133% |
| qwen2 3B Q4_0_8_8              |   1.69 GiB |     3.09 B | CPU        |      64 |        pp1024 |       279.86 ± 45.63 |          100% |
| qwen2 3B Q4_0_8_8              |   1.69 GiB |     3.09 B | CPU        |      64 |        pp2048 |        320.77 ± 5.00 |          124% |
| qwen2 3B Q4_0_8_8              |   1.69 GiB |     3.09 B | CPU        |      64 |         tg128 |         43.51 ± 0.05 |          111% |
| qwen2 3B Q4_0_8_8              |   1.69 GiB |     3.09 B | CPU        |      64 |         tg256 |         43.35 ± 0.09 |          110% |
| qwen2 3B Q4_0_8_8              |   1.69 GiB |     3.09 B | CPU        |      64 |         tg512 |         42.60 ± 0.31 |          105% |

Q4_0_8_8 offers a nice bump to prompt processing and a small bump to text generation

Which File to Choose?

Click here for details

A great write - up with charts showing various performances is provided by Artefact2 here

The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.

If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1 - 2GB smaller than your GPU's total VRAM.

If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, and aim for a quant with a file size 1 - 2GB smaller than that total.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご