[

Ornstein-3.6-27B-RYS

RYS-enhanced variant of the Ornstein-3.6-27B dense model. Layer 33 is duplicated using the Repeat Your Self (RYS) method, improving reasoning and instruction-following performance without increasing active parameter count at inference time.

GGUF quantizations: GestaltLabs/Ornstein-3.6-27B-RYS-GGUF

About Gestalt Lab

We are a proudly Canadian research collective working to advance sovereign Canadian AI — open-weight models that Canadians (and everyone else) can run locally, study, and build on without dependence on closed foreign APIs. All training, fine-tuning, and quantization is done on local and self-funded compute. By supporting this work, you help keep frontier model development accessible, transparent, and under Canadian stewardship.

Important: requires a patched llama.cpp

RYS duplicates one of the middle layers, which breaks the hardcoded full_attention_interval = 4 assumption in stock llama.cpp's Qwen3.5 loader. This model is converted with per-layer head_count_kv baked in, and you need a llama.cpp that reads that per-layer metadata instead of falling back to the interval formula.

Patched fork: https://github.com/DJLougen/llama.cpp (default branch rys-qwen35, fully backward-compatible).

Stock llama.cpp, Ollama, LM Studio, and any other inference runtime built on stock llama.cpp will currently fail to load this model with a check_tensor_dims error — this is expected until/unless the patch is upstreamed.

Support This Work

Our training compute is entirely self-funded. If this model is useful to you, consider supporting the lab:

Support on Ko-fi

Model Details

Architecture: Qwen3.5 dense
Parameters: ~27B active
Layers: 65 (64 original + 1 RYS-duplicated layer 33)
Context length: 131,072 tokens
License: Apache-2.0

Usage

Build the patched llama.cpp

git clone https://github.com/DJLougen/llama.cpp.git
cd llama.cpp
git checkout rys-qwen35
cmake -B build -DGGML_CUDA=ON -DCMAKE_BUILD_TYPE=Release
cmake --build build -j

Drop -DGGML_CUDA=ON for a CPU-only build. The patch touches the GGUF loader; backend selection is independent.

Download + run

./build/bin/llama-server \
    -m ornstein-3.6-27b-rys-q4_k_m.gguf \
    --host 0.0.0.0 --port 8080 \
    --n-gpu-layers 99 --ctx-size 131072 \
    --flash-attn on --jinja \
    -ctk q4_0 -ctv q4_0