Interim Labs Huihui Qwen3.5-4B abliterated · GGUF Q4_K_M
A standalone GGUF Q4_K_M release published by Interim Labs for local text generation. This package derives from huihui-ai/Huihui-Qwen3.5-4B-abliterated. The GGUF weights are redistributed byte-for-byte from the credited conversion source below.
Model series
Part of the Interim Labs Qwen3.5 series: three sizes, each with an original Qwen3.5 variant and a Huihui abliterated variant, in two formats.
| Model | MLX 4-bit | GGUF Q4_K_M |
|---|---|---|
| Qwen3.5 2B | MLX | GGUF |
| Huihui Qwen3.5 2B abliterated | MLX | GGUF |
| Qwen3.5 4B | MLX | GGUF |
| Huihui Qwen3.5 4B abliterated | MLX | GGUF |
| Qwen3.5 9B | MLX | GGUF |
| Huihui Qwen3.5 9B abliterated | MLX | GGUF |
Package details
| Property | Value |
|---|---|
| Parameter class | 4B |
| Format | GGUF |
| Quantization | Q4_K_M |
| Weight file | model.gguf |
| Weight size | 2.71 GB (2,707,514,688 bytes) |
| GGUF architecture | qwen35 |
| Context in GGUF metadata | 262,144 tokens; practical limits depend on runtime and memory |
No tensors were rewritten in this release. The context value in the GGUF metadata is not a claim of tested long-context quality or of memory requirements.
Usage and runtime requirements
Download the package with the Hugging Face CLI:
hf download interimlabs/InterimLabs-Huihui-Qwen3.5-4B-abliterated-GGUF-Q4_K_M model.gguf
Load model.gguf in a GGUF runtime with Qwen3.5 (qwen35) support, such as a compatible build of llama.cpp. Q4_K_M support alone is insufficient if the runtime does not support the model architecture. Choose context and GPU offload settings for your available memory. Compatibility across applications has not been comprehensively verified.
Sources and conversion
- GGUF source: mradermacher/Huihui-Qwen3.5-4B-abliterated-GGUF at
4a5daa6fbefca5fe822dc65fcb95cc4576fa9720. - Source weight file:
Huihui-Qwen3.5-4B-abliterated.Q4_K_M.gguf. - Reference upstream: huihui-ai/Huihui-Qwen3.5-4B-abliterated at
5581467dfd52bf338c782006a6cdce05c42594be.
The conversion source does not establish the exact upstream revision used for its conversion. The reference revision above documents the inspected upstream snapshot, not a verified conversion lineage.
See provenance.json for source and package metadata.
Weight-file SHA-256:
423f10b6ec2d99c3378143d7cd3b80eb4887b3ed92103103ac59173b404f4f7c
Evaluation and limitations
This release does not include a comprehensive benchmark of the packaged model. No writing-quality, reliability, or performance improvement over the credited source is claimed.
The Huihui variant uses an upstream abliterated checkpoint. This release does not establish a measured refusal-rate improvement, and abliteration should not be taken as a guarantee of answer quality.
A native Linux/Windows application installation test was not completed in the original package evaluation.
License and credits
Distributed under Apache-2.0. Credit belongs to the Qwen team, Huihui, and the conversion authors linked above. Interim Labs publishes this package and its documentation. See the linked upstream cards for the original model details and limitations.