Qwen3.5-2B-CyberSec — GGUF

GGUF exports of reaperdoesntknow/Qwen3.5-2B-CyberSec for local inference with llama.cpp-compatible runtimes.

The source model is associated with the Trendyol Cybersecurity Instruction Tuning Dataset. No benchmark or safety-evaluation results are published with this GGUF release.

Available files

File	Reported size	Practical use
`Qwen3.5-2B.BF16.gguf`	3.90 GB	Highest-fidelity GGUF build
`Qwen3.5-2B.Q8_0.gguf`	2.08 GB	High-quality quantized build
`Qwen3.5-2B.Q4_K_M.gguf`	1.31 GB	Recommended local starting point
`Qwen3.5-2B.BF16-mmproj.gguf`	repository artifact	Multimodal projector; see caveat below

Run locally

llama-cli \
  -hf reaperdoesntknow/Qwen3.5-2B-CyberSec-GGUF:Q4_K_M \
  --jinja

With Ollama:

ollama run hf.co/reaperdoesntknow/Qwen3.5-2B-CyberSec-GGUF:Q4_K_M

Multimodal note

A BF16 projector file is present, and the source-model configuration includes vision components. This release does not include a documented multimodal smoke-test receipt. Verify the projector, prompt format, runtime version, and image path before claiming multimodal support.

Intended use

Local qualitative evaluation of the source checkpoint.
CPU or consumer-GPU experimentation.
Comparison of BF16, Q8_0, and Q4_K_M output quality.

Limitations and safety

Quantization can change quality and behavior relative to the source checkpoint.
The release has no published benchmark, regression, or safety report.
Generated security instructions may be wrong, unsafe, or dual use.
Do not execute generated commands without review and isolation.
Do not rely on the model as an authoritative security control or decision-maker.

Lineage and license

Source checkpoint: reaperdoesntknow/Qwen3.5-2B-CyberSec
Recorded upstream base: unsloth/Qwen3.5-2B
Recorded dataset: Trendyol/Trendyol-Cybersecurity-Instruction-Tuning-Dataset
Format: GGUF
License: Apache-2.0

Part of the CIx cybersecurity model collection.