OFFSITE.DARK
← Greps

Jul 5, 2026

1 min

Uploaded by hotdogs

HF Model
hotdogs/qwen3.6-27b-cybersecurity-lora
Base Model
Qwen/Qwen3.6-27B
Pipeline
text-generation
Downloads
517
Likes
9
  • cybersecurity
  • lora
  • qwen3.6
  • sft
  • fenrir
  • peft

cyber-llm

Community upload on Hugging Face — indexed for inquiry, not endorsement.

Qwen3.6-27B Cybersecurity LoRA (Fenrir SFT)

Uploaded to Hugging Face by hotdogs — PEFT LoRA on Qwen3.6-27B trained with AlicanKiraz0 Fenrir cybersecurity dataset; GGUF tags present.

What Is This?

Community upload on Hugging Face by hotdogs. qwen3.6-27b-cybersecurity-lora is a PEFT LoRA adapter on Qwen/Qwen3.6-27B, supervised-fine-tuned with the AlicanKiraz0/Cybersecurity-Dataset-Fenrir-v2.1 corpus (same Fenrir lineage as other cyber LLMs already on Greps). Tags include gguf for local inference packaging. It is not an OFFSITE.DARK release.

Created mid-June 2026; last modified 2026-07-05 in this indexing window.

Metadata

FieldValue
Authorhotdogs
Base modelQwen/Qwen3.6-27B
MethodLoRA / SFT (PEFT)
DatasetAlicanKiraz0/Cybersecurity-Dataset-Fenrir-v2.1
Pipelinetext-generation
LicenseApache-2.0
Downloads~517
Likes9

Why Index It

  • Continues the Fenrir/Qwen cybersecurity fine-tune cluster already represented by Titus / BaronLLM-adjacent greps.
  • 27B class is large enough for serious local cyber-assistant experiments.
  • Useful contrast to smaller Strix XSS / RavenX agent greps.

Caveats

LoRA adapters inherit base-model jailbreak and hallucination risks. Use only in authorized research sandboxes; do not treat outputs as authoritative vuln intel.

Sources