---

license: apache-2.0

base_model: Jackrong/Qwopus3.6-27B-Coder

---

⚠️ THIS REPO HAS BEEN RENAMED

This repository is deprecated. The contents have been moved to:

Fredred89/Qwopus3.6-27B-Coder-GGUF-kaitchup-MoQ-4.0

Why the rename: The "Predator-Q" branding implied novel work, but the underlying GGUF is a direct application of kaitchup's MoQ recipe (from kaitchup/Qwen3.6-27B-GGUF-MoQ) to the Qwopus3.6-27B-Coder model. The new repo name properly attributes the source.

What we actually did:

1. Converted Qwopus3.6-27B-Coder from safetensors → F16 GGUF (~30 min, 53.8 GB)
2. Generated an importance matrix via llama-imatrix (~1 hour)
3. Applied kaitchup's MoQ recipe at 4.0 BPW via llama-quantize (~5 min)
4. Validated with LCB-30 (LiveCodeBench easy subset, 30 problems)

The actual GGUF file is unchanged (same SHA256: 587840e75895199e5ad771bfa7dfd9682f6d85ae295ad00001b78adb485c52c1). It just has a properly attributed name now.

---

Attribution

• Recipe source: kaitchup/Qwen3.6-27B-GGUF-MoQ by Benjamin Marie (kaitchup)
• Methodology: MoQ (Mixture of Quantizations) by Waleed Ahmad (w-ahmad)
• Base model: Jackrong/Qwopus3.6-27B-Coder
• Quantization tool: llama.cpp (open source)

The new repo (Fredred89/Qwopus3.6-27B-Coder-GGUF-kaitchup-MoQ-4.0) contains the same GGUF plus multi-benchmark validation results (HumanEval+ 164, MBPP+ 100, BigCodeBench 50, LCB-30 30).