This model makes no sense
Is this a finetuned model? If so why is it catagorized as a quantization of (DavidAU/Qwen3.6-27B-Heretic2-Uncensored-Finetune-Thinking).
Did you just rename their model after quantizing it and not change anything? Because all that does is confuse people as to the purpose of your model.
Please clarify this issue
Hey rombodawg — fair to raise the question, so let me be transparent about what actually happened here.
This is a fine-tune, not a rename. The "quantization of" tag in the HuggingFace metadata is auto-generated based on the base_model field — it doesn't capture the full lineage. Here's the actual pipeline:
Base selection — We chose DavidAU/Qwen3.6-27B-Heretic2-Uncensored-Finetune-Thinking specifically because the community was asking for Qwen3 architecture (see discussion #1). Starting from an already-abliterated base means we're not wasting compute re-teaching it to be uncensored.
Custom dataset — 2,541 training examples curated from: HackerOne disclosed reports, CVE writeups, Bug Bounty Daily CoT chains, GitHub security research (2024–2026). Not publicly available as a single dataset anywhere. This took weeks of ETL pipelines, scraping, normalization, and DPO-format conversion (including a → tag migration for Qwen3's native vocab).
Training — SFT via Unsloth on a RunPod H100 80GB, LoRA r=16 injected across all 7 attention + MLP modules across 64 transformer layers, 2 epochs. The VRAM math alone required significant engineering work (54GB base model + optimizer states).
Merge + IMatrix Quant — LoRA merged back to BF16 full weights, then quantized with IMatrix guidance via llama.cpp to Q4_K_S and Q6_K.
Benchmarked — 5/5 on our BugTraceAI Ultra Bench v1.0 (Nuclei templates, CVE PoCs, JWT cracking, kernel exploits) at 0% refusal rate.
All of this is already documented in the model card under 🧠 Training Details — worth a read before assuming. The base model is the foundation, not the product. The work is in the data, the training objective, and what the model now does that the base didn't.
Can we possibly have the q8 quantized version as well? q6 seems to be hallucinating and jumping to conclusions. It falsely identified cloakbrowser as fully open source when it's only half open source and the actual browser build is closed source proprietary . Overall it seems weaker than the q8 qwen 27B I've been using. It degraded almost to ornith 35B MoE level where it had only shallow reasoning and didnt look deep into things
(Btw I'm sure nobody wants to hear this, but as far as "consumer grade GPUs" are concerned, an a6000 with 48GB of VRAM now counts as high end consumer GPU, a 4090 with 24GB of VRAM is considered entry level, and anything below that... well.... it's not pretty. What is a single rtx 6000 pro considered? Something like a gaming rig or stuff for creators to use. Now, only MULTIPLE rtx 6000 pros are considered "out there". I say this because I learned the hard way that the minimum quality a gguf can have is q8 to retain some semblance of the awesomeness of the full weights... )
Thanks for the feedback!
To clarify, BugTraceAI-CORE-Ultra is a highly specialized SFT/DPO model designed specifically for security tooling, exploit generation, and action automation (e.g., generating Nuclei templates, CVE PoCs, and structured pentesting scripts). It is not optimized to act as a general-knowledge or deep reasoning model (such as answering specific factual questions about third-party software licensing).
For deep security reasoning, threat modeling, and chain-of-thought analysis, we have the Apex series, which is built exactly for those tasks. You can find them here:
Full Weights (FP16):
https://huggingface.co/BugTraceAI/BugTraceAI-Apex-G4-26B-Master-f16
Quantized (Q4): BugTraceAI-Apex-G4-26B-Q4
https://huggingface.co/BugTraceAI/BugTraceAI-Apex-G4-26B-Q4
Regarding the quantization on this model: with modern IMatrix (Importance Matrix) calibration, the difference in perplexity between Q6_K and Q8_K is negligible, especially for highly structured code syntax. The hallucination you encountered is a knowledge-cutoff or dataset-scope limitation, not a quantization loss (even at Q8 or FP16, this SFT model would likely yield similar results for that specific general-knowledge prompt).
Letme share this article with you.
https://albert-corzo.medium.com/theyre-selling-you-hype-90-of-people-are-using-ai-models-wrong-and-don-t-even-know-it-74dbaf8e95a6
That said, we want to accommodate everyone's setup, so I will queue a Q8_K quant upload for this CORE-Ultra model shortly so you can test it on your hardware.
Stay tuned!