🪄 Qwen3-8B-Houdini-VEX-v10 (Grandmaster Edition)

[!TIP] 🚀 Flagship Release Available: Explore the new Qwen3-8B-Houdini-VEX-v11 (Grandmaster Edition) featuring Compiler-in-the-Loop (CITL) validation, Q8_0 quantizations, and reinforced dual-mode reasoning!

Qwen3-8B-Houdini-VEX-v10 is a state-of-the-art neural code intelligence model fine-tuned specifically for SideFX Houdini VEX. Built on top of Qwen3-8B, it is engineered to understand 3D vector spaces, quaternion orientations, SIMD CVEX parallelism, geometry attribute classes, and procedural graphics algorithms.

Developed by Anshul Vashist.

🚀 Key Highlights & Capabilities

• Dual-Mode Inference:
  • ⚡ Turbo Mode: Direct, instant VEX snippets for rapid procedural modeling with zero conversational filler.
  • 🧠 Deep Reasoning Mode: Activates explicit <think>...</think> architectural analysis, solving 3D vector trigonometry, spatial hash lookups, and SIMD mechanics before emitting code.
• Mastered Contexts:
  • Point Wrangling: Stream processing (@P, @N, @Cd, @v, @Time).
  • Vertex Wrangling: UV layout, tangent space, and mesh connectivity (v@uv, @vtxnum).
  • Primitive Wrangling: Face intrinsic attributes, volume voxels, and packed transforms.
  • Detail Wrangling: Global procedural generators, addpoint(), addprim(), and parameter hoisting.
• Advanced Ecosystems:
  • KineFX: Skeleton rigging, joint transforms (4@localtransform), 2-bone IK solvers, and lookat constraints.
  • Solaris / USD: Stage traversal, prim relationships, and USD attributes (usd_attrib, usd_setattrib).
  • Pyro & Volumetrics: Divergence-free curl noise, thermal buoyancy, and physically accurate blackbody(temp, intensity).
  • Negative Error Repairs: Pre-trained to repair broken GLSL habits (vec3 -> set), quaternion semantics (p@orient vs v@orient), and VEX syntax (smooth vs smoothstep).

📦 Model Files in this Repository

💻 How to Use

Option 1: Standalone llama-server (Zero Dependencies)

Run the base model with the LoRA adapter locally using llama-server:

llama-server.exe \
  --model Qwen3-8B-Houdini-VEX-v10-Q5_K_M.gguf \
  --lora qwen3-vex-v10-lora.gguf \
  --port 58421 \
  --host 127.0.0.1 \
  --ctx-size 4096 \
  --n-gpu-layers 35

Option 2: Live in SideFX Houdini (HDA)

The custom i_attribwrangle.hda digital asset, GUI installers, and Python controllers are maintained directly on GitHub: 👉 Download HDA & Installer: https://github.com/anshul-vashist/Houdini-Ai-Attribute-Wrangle

1. Clone or download the GitHub repository.
2. Place the model weights (Qwen3-8B-Houdini-VEX-v10-Q5_K_M.gguf or base + qwen3-vex-v10-lora.gguf) in the models/ folder.
3. Use the 1-click GUI installer (Setup_AI_Wrangle.exe) or drag install_in_houdini.py directly into Houdini!

⚡ Prompting Format

Turbo Fast Mode (Instant VEX):

<|im_start|>system
You are an expert Houdini VEX compiler. Output ONLY pure VEX snippet code. No markdown, no backticks, no explanations, no HTML tags.<|im_end|>
<|im_start|>user
Context: Point wrangle
Task: Displace points along normal with 3D curl noise
Write pure Houdini VEX code:<|im_end|>
<|im_start|>assistant

Deep Reasoning Mode:

<|im_start|>system
You are Houdini VEX Copilot, an expert AI programming assistant fine-tuned and created by Anshul Vashist.
When presented with a task, FIRST think through the problem in an explicit <think>...</think> block:
  1. Analyze input geometry, vector spaces, and physical/mathematical equations.
  2. Outline algorithmic steps and procedural logic (loops, spatial lookups, matrix transforms).
AFTER the </think> tag, output ONLY the pure, verified Houdini VEX code.<|im_end|>
<|im_start|>user
Context: Detail wrangle
Task: Generate a Fibonacci spiral distribution of 100 points on a unit sphere using golden ratio angle
Write pure Houdini VEX code:<|im_end|>
<|im_start|>assistant
<think>

📜 License

Apache-2.0. Free for commercial and personal procedural graphics production.