Add 3D Image Generation Model (shap-E) & DeepCoder-1.5 B

[eskayML]

Template 1 - shap E 3D model

This is a very simple Nosana AI template that converts text descriptions into 3D models using OpenAI's Shap-E model.

https://github.com/openai/shap-e/

Features

• Text-to-3D generation with detailed controls
• Animated GIF output showing the 3D model from multiple angles
• Configurable parameters for quality and style control
• GPU acceleration for fast generation (min 16GB OF VRAM)

Notes:

• Made use of huggingface-cli download to forcefully download the weights to ensure that the weights are properly downloaded when running the nosana instance ( due to some incompatibility issues with downloading the weights directly from the hub inside the code).
• When gpu is set to false , easily switch the app.py to the client_app.py inside the job-definition.json , to make use of inference api.

Nosana Run

JOB URL: https://dashboard.nosana.com/jobs/CJez3fP7LYeNjLnG9ybezVALoqdTau9PN2kYUVXbJHkQ

---

Template 2 - DeepCoder 1.5B

This template deploys a high-performance code generation model (DeepCoder-1.5B) with an OpenAI-compatible API endpoint.

Key Components:

• Uses vLLM for optimized inference
• Exposes port 9000 for API access
• Requires 8GB VRAM for operation
• Provides multi-language code generation capabilities

Technical Details:

• Model: agentica-org/DeepCoder-1.5B-Preview
• API format: OpenAI-compatible

Nosana Run

JOB URL: https://dashboard.nosana.com/jobs/89JkqUAY3y1MULRp7QLRovguDad169CgBkfiVPSBKvxt