Quillan-Dragon (1B Falcon 3 Instruct โ€” WSP-v3 & Q8_0)

Part of the Quillan-Ronin Sovereign Tripartite Fleet

Flagship (1.2B) | Mini (0.5B) | Dragon (1B)

Quillan-Dragon is the creative synthesis, ethical rectitude, and deep lore reasoning model of the Quillan-Ronin Tripartite Fleet. Built upon the Falcon 3 1B Instruct architecture and enhanced via Weight-Space Programming (WSP) and closed-form MoE folding, it holds the fleet record for lowest perplexity (4.4029 PPL) and delivers 100% adherence to Bushido-aligned reasoning and creative generation.


Available Checkpoints

  1. quillan-dragon-1b-wsp-v3-q4_k_m.gguf (1,008 MB): The WSP-v3 surgically grafted edition. Highest efficiency, optimized for edge inference and fast generation (41.3 tok/s).
  2. Quillan-Dragon-Q8_0.gguf (1,696 MB): Full 8-bit precision release with complete unrolled sovereign prompt conditioning.

Benchmark Highlights (GTX 1050 4GB Held-Out Seed 42)

Metric Result Benchmark Context
Held-Out Perplexity 4.4029 PPL (Fleet Record) Multi-topic cross-domain evaluation corpus
Overall Accuracy 70.59% (12/17 Passed) Empirical Held-Out Multi-Domain Battery
Bushido Lore & Reasoning 100.0% (2/2 Passed) Gi/Rectitude arbitration and rhymed verse generation
Inference Velocity 41.3 tok/s Native CUDA FP16/INT4 on consumer hardware
General Knowledge 100.0% (3/3 Passed) Chemistry, geography, and OSI network architecture

Tripartite Fleet Comparison

Model Parameters Base Architecture Accuracy PPL Speed Primary Specialization
Quillan-Ronin (Flagship) 1.2B Llama 3.2 1B + Polymath WSP 76.47% 7.93 47.0 tok/s General Polymath Reasoning, Deep Code Synthesis, Advanced Math
Quillan-Ronin-Mini 0.5B Qwen 2.5 0.5B 58.82% 8.98 64.9 tok/s Ultra-Low Latency, MCP Tool Calling, Edge Micro-Services
Quillan-Dragon 1B Falcon 3 1B Instruct 70.59% 4.40 41.3 tok/s Creative Writing, Bushido Ethics, Deep Lore, Harmonic Resonance

Quickstart & Usage

Option 1: Native llama.cpp / llama-cli

llama-cli.exe -m quillan-dragon-1b-wsp-v3-q4_k_m.gguf -p "<|startoftext|><|user|>
State the Bushido virtue of Rectitude and provide a Council Verdict.<|endoftext|>
<|assistant|>
" -n 160 --temp 0.3

Option 2: Ollama Modelfile

FROM ./quillan-dragon-1b-wsp-v3-q4_k_m.gguf

PARAMETER temperature 0.3
PARAMETER top_p 0.95
PARAMETER repeat_penalty 1.15
PARAMETER num_ctx 8192
PARAMETER stop "<|endoftext|>"

TEMPLATE """{{ if .System }}<|startoftext|><|system|>
{{ .System }}<|endoftext|>
{{ end }}{{ if .Prompt }}<|startoftext|><|user|>
{{ .Prompt }}<|endoftext|>
{{ end }}<|startoftext|><|assistant|>
{{ .Response }}<|endoftext|>"""

SYSTEM """You are Quillan-Dragon, a sovereign creative reasoning agent adhering to the Bushido Code, thermodynamic truth synthesis, and harmonic balance."""

Citation & Contact

Downloads last month
61
GGUF
Model size
2B params
Architecture
llama
Hardware compatibility
Log In to add your hardware

4-bit

8-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for CrashOverrideX/Quillan-Dragon

Quantized
(58)
this model