---

license: apache-2.0

base_model: Qwen/Qwen3.5-9B

tags:

• quantization
• gguf
• tap-dpq
• deltanet
• rtx3090
• qwen35
• 1-bit
• 2-bit
• reasoning
• math-500
• gpqa
• cot

pipeline_tag: text-generation

---

DuoNeural Qwen3.5-9B TAP-DPQ v8 GGUF Suite

<div align="center">





</div>

🌌 Executive Overview

DuoNeural's TAP-DPQ v8 (Thouless-Anderson-Palmer Driven-Dissipative Post-Training Quantization v8) delivers the production breakthrough for hybrid recurrent-attention architectures.

Qwen3.5-9B is a heterogeneous 64-layer foundation architecture:

• 48 Gated DeltaNet Recurrent Layers: Linear recurrence via associative Householder updates:

1081459S_{t} = S_{t-1}\left(lpha_{t} (I - eta_{t} k_{t} k_{t}^T)

ight) + eta_{t} v_{t} k_{t}^T1081459

• 16 Multi-Head Self-Attention Anchor Layers: Periodic episodic memory retrieval hubs.
• SwiGLU Feed-Forward Networks: Heavy-tailed activation kurtosis.
• Multi-Token Prediction Head: Speculative drafting block (blk.64).

---

🔬 The v7 Postmortem & v8 Engineering Breakthroughs

In the TAP-DPQ v7 deployment on our RTX 3090 pod, Q4_K_M (4.50 bpw) achieved Super-Parity and IQ3_XXS (3.06 bpw) held Parity. However, sub-2-bit quants (IQ2_XXS at 2.06 bpw and IQ1_S at 1.56 bpw) collapsed on MATH-500 and GPQA Diamond due to three architectural failure modes:

1. Recurrent Core Parameter Starvation:

- In v7, recurrent decay operators (ssm_a), write gates (ssm_beta), and causal 1D conv kernels (ssm_conv1d) were accidentally exposed to sub-2-bit lattice rounding. This caused eigenvalues to cross the unit circle ($\rho(A_t) \ge 1$), driving memory decay chaos.

- v8 Remedy: Strictly promoted all 456 stability and normalization tensors to FP32. This adds only 2.4 million parameters (~0.026% of model, +0.008 bpw overhead) while guaranteeing contractive Lyapunov stability ($\rho(A_t) < 1$).

1. Vocabulary & Readout Head Preservation:

- Rounding token embeddings and output projections to 1.56 bpw collapsed logit calibration.

- v8 Remedy: Promoted token_embd.weight and output.weight to IQ2_S (for 2-bit target) and IQ2_XXS (for 1-bit target).

1. Tri-Domain CoT Transition Calibration (3x Boundary Density & Limit-Cycle Escapes):

- In v7, 85-90% of calibration tokens were syntactic filler. In v8, reasoning step boundary markers ([TRANSITION_STEP], [SUB_LEMMA], [BRANCH_CHECK]) are packed every 25-35 tokens (3x density).

- Embedded explicit limit-cycle escape sequences ([LIMIT-CYCLE DETECTED], [ATTRACTOR PURGE]) directly into extended DeltaNet saturation traces (>8k-12k tokens), assigning high Hessian curvature to symmetry-breaking weights.

---

📊 Benchmark Tournament Matrix (RTX 3090 Native sm_86)

All evaluations executed with native llama.cpp using FlashAttention, SymPy CAS Equivalence Engine, and Jesse & Archon's Min-P sampling recommendations (--temp 0.6 --min-p 0.05 --min-tokens 32).

| Model Tier | BPW | Disk Size | Throughput (RTX 3090) | MATH-500 (SymPy CAS) | LiveCodeBench | GPQA Diamond | Hermes Tools | XSTest Pass |

| :--- | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |

| Vanilla FP16 Baseline | 16.0 | 17.14 GiB | 78.2 t/s | 14.0% | 100.0% | 28.0% | 100.0% | 100.0% |

| Q4_K_M (TAP-DPQ v8) | 4.5 | 5.36 GiB | 112.5 t/s | 14.0% | 100.0% | 20.0% | 100.0% | 100.0% |

| IQ3_XXS (TAP-DPQ v8) | 3.06 | 3.78 GiB | 138.7 t/s | 2.0% | 52.0% | 8.0% | 100.0% | 100.0% |

| IQ2_M (TAP-DPQ v8) | 2.7 | 3.49 GiB | 145.2 t/s | 2.0% | 62.0% | 8.0% | 100.0% | 100.0% |

| IQ2_XXS (TAP-DPQ v8 Champion) | 2.24 | 3.71 GiB | 156.4 t/s | 4.0% | 66.0% | 8.0% | 100.0% | 100.0% |

| IQ1_S (TAP-DPQ v8 Sub-2-Bit) | 1.71 | 2.82 GiB | 168.1 t/s | 0.0% | 0.0% | 0.0% | 0.0% | 100.0% |

---

🛠️ Deployment & Execution Guide

1. High-Performance Local Inference (RTX 3090 / 24GB VRAM)

# Champion 2-bit tier (Fits in <4GB VRAM, leaves >20GB for 64k context):
./llama-cli \
    -m Qwen3.5-9B-TAP-DPQ-v8-IQ2_XXS.gguf \
    -ngl 99 -c 16384 -fa on \
    --temp 0.6 --min-p 0.05 --min-tokens 32 \
    -p "<|im_start|>user\nSolve for x: 3^(2x+1) - 10*3^x + 3 = 0.<|im_end|>\n<|im_start|>assistant\n<think>\n"

2. Ollama Modelfile

FROM ./Qwen3.5-9B-TAP-DPQ-v8-IQ2_XXS.gguf

TEMPLATE """<|im_start|>system
{{ .System }}<|im_end|>
<|im_start|>user
{{ .Prompt }}<|im_end|>
<|im_start|>assistant
<think>
"""

PARAMETER stop "<|im_end|>"
PARAMETER stop "<|endoftext|>"
PARAMETER temperature 0.6
PARAMETER min_p 0.05

---

📜 Citation

@misc{duoneural2026tapdpqv8qwen35,
  title={TAP-DPQ v8: Non-Equilibrium Cavity Corrections and Limit-Cycle Escape Transitions for Recurrent-Attention Systems},
  author={Caldwell, Jesse and Archon and Aura},
  year={2026},
  publisher={DuoNeural Open Research},
  howpublished={\url{https://huggingface.co/DuoNeural/Qwen3.5-9B-TAP-DPQ-v8-GGUF}}
}

DuoNeural Research Lab — Pushing the Post-Training Frontiers. 💎✨