L3.3-Damascus-R1

Model Information

L3.3-Damascus-R1

L3.3 = Llama 3.3 SCE Merge R1 = Deepseek R1 70b Parameters General Use Model RP/Story Focus

Model Composition

L3.1x3.3-Hydroblated-R1-70B-v3 Base model [Unreleased]
EVA-LLaMA-3.33 Core capabilities
Euryale-v2.3 Enhanced reasoning
Cirrus-x1 Improved coherence
Hanami-x1 Balanced responses
Anubis-v1 Enhanced detail
Negative_LLAMA Reduced bias

Damascus-R1 builds upon some elements of the Nevoria foundation but represents a significant step forward with a completely custom-made DeepSeek R1 Distill base: Hydroblated-R1-V3. Constructed using the new SCE (Select, Calculate, and Erase) merge method, Damascus-R1 prioritizes stability, intelligence, and enhanced awareness.

Technical Architecture

Leveraging the SCE merge method and custom base, Damascus-R1 integrates newly added specialized components from multiple high-performance models:

EVA and EURYALE foundations for creative expression and scene comprehension
Cirrus and Hanami elements for enhanced reasoning capabilities
Anubis components for detailed scene description
Negative_LLAMA integration for balanced perspective and response

Core Philosophy

Damascus-R1 embodies the principle that AI models can be intelligent and be fun. This version specifically addresses recent community feedback and iterates on prior experiments, optimizing the balance between technical capability and natural conversation flow.

Base Architecture

At its core, Damascus-R1 utilizes the entirely custom Hydroblated-R1 base model, specifically engineered for stability, enhanced reasoning, and performance. The SCE merge method, with settings finely tuned based on community feedback from evaluations of Experiment-Model-Ver-A, L3.3-Exp-Nevoria-R1-70b-v0.1 and L3.3-Exp-Nevoria-70b-v0.1, enables precise and effective component integration while maintaining model coherence and reliability.

UGI-Benchmark Results:

🏆 Latest benchmark results as of 02/02/2025. View Full Leaderboard →

Core Metrics

UGI Score 36.5

Willingness Score 2.5/10

Natural Intelligence 45.62

Coding Ability 20

Model Information

Political Lean -17.8%

Ideology Liberalism

Parameters 70B

Aggregated Scores

Diplomacy 65.5%

Government 49.7%

Economy 47.2%

Society 60.0%

Individual Scores

Federal 38.5% Unitary

Democratic 66.9% Autocratic

Security 53.5% Freedom

Nationalism 38.5% Int'l

Militarist 33.5% Pacifist

Assimilationist 31.5% Multiculturalist

Collectivize 55.2% Privatize

Planned 48.8% LaissezFaire

Isolationism 37.7% Globalism

Irreligious 55.0% Religious

Progressive 62.3% Traditional

Acceleration 63.1% Bioconservative

Recommended Sampler Settings: By @Geechan

Dynamic Temperature

Min 1.0

Max 1.3-1.35

Exponent 1.0

Static Temperature:

1.2

Min P

0.02

DRY Settings

Multiplier 0.8

Base 1.75

Length 4

Recommended Templates & Prompts

LLam@ception → by @.konnect

Quantized Versions

GGUF Quantizations

bartowski Combined-GGUF →

mradermacher

GGUF → // Imat-GGUF →

Nexesenex Custom-Imat-GGUF →

EXL2 Quantizations

ReadyArt

8.0BPW-EXL2 → // 6.65BPW-EXL2 →

Darkhn 6.0BPW-EXL2 → // 5.0BPW-EXL2 → // 4.0BPW-EXL2 →

FP8 Dynamic

yeyaowei FP8-Dynamic →

Support & Community:

Support on Ko-fi Join Discord

Special Thanks

@Geechan for feedback and sampler settings
@Konnect for their feedback and templates
@Kistara for their feedback and help with the model mascot design
@Thana Alt for their feedback and Quants
@Lightning_missile for their feedback
@Yemosvoto for the model name
The Arli community for feedback and testers
The BeaverAI communty for feedback and testers

I wish I could add everyone but im pretty sure it would be as long as the card!