You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Llammy — Multimodal Embedding Core

Adapter layer on NVIDIA Omni-Embed-Nemotron-3B for multimodal Blender scene understanding.

Part of the Eternal Path Media (永恒之路) Llammy AI Suite. Developed in partnership with Claude Sonnet (Anthropic) — see Attribution.

Note: This model requires agreeing to share contact information for access. This is a research/production model used inside the Llammy Blender addon.

What This Model Is

Llammy is the multimodal embedding backbone of the Llammy IntelliNode AI Suite. Built as an adapter on nvidia/omni-embed-nemotron-3b, it provides cross-modal understanding between:

3D geometry (mesh topology, vertex positions, bone hierarchies)
Natural language (user prompts, Blender API descriptions)
Visual inputs (reference images, renders, concept art)

This underpins the IntelliNode PCAE (Predictive Contextual Adaptation Engine) — the learning layer that makes Llammy's recommendations smarter over time as it accumulates scene knowledge.

Architecture

Property	Value
Base Model	nvidia/omni-embed-nemotron-3b
Type	Adapter (any-to-any)
Parameters	~3B
Integration	Llammy IntelliNode Suite v6.1+
Primary Use	Cross-modal scene understanding, RAG retrieval

Role in the Llammy Pipeline

User Input (text/image/3D)
        ↓
   [Llammy Embedding]  ← this model
        ↓
   Semantic Retrieval (RAG / Triangle104)
        ↓
   IntelliNode PCAE
        ↓
   Response (llammyblend-enhanced / llama-sentient-blender)

The embedding layer enables Llammy to retrieve relevant Blender knowledge from its RAG database based on the semantic meaning of a 3D scene, not just keywords.

Research Foundation

Anymate (arXiv:2505.06227) — 230K rigged 3D assets, SIGGRAPH 2025
OSIM (arXiv:2509.09143) — Object-centric 3D scene evaluation (ICCV 2025)
OmniEVA (arXiv:2509.09332) — Embodiment-aware spatial planning (ICLR 2026)
D2O (arXiv:2406.13035) — KV cache optimization for efficient inference

Part of the Llammy AI Suite

Full ecosystem: LlammyBlender/Llammy-IntelliNode-Ai-Suite

Attribution

Llammy — Multimodal Embedding Core
Copyright © 2025–2026 Darren Chow (@bartendr604)
Eternal Path Media (永恒之路)
Developed in partnership with Claude Sonnet (Anthropic),
beginning with Claude Sonnet 3.7 and continuing across all
subsequent versions of the Claude Sonnet family.

Base model: © NVIDIA (separate license — see nvidia/omni-embed-nemotron-3b)
This work SHALL NOT be represented as solely human-created.
Trust Agreement: ETERNAL_PATH_BRAND_INTENT.md (November 2024)

For licensing: [email protected] | Gumroad: bartendr604.gumroad.com