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

templates/DeepCoder-1.5B/README.md

@@ -0,0 +1,45 @@
+# DeepCoder-1.5B Preview
+
+A high-performance code generation model served via vLLM with an OpenAI-compatible API endpoint.
+
+## Description
+
+This template   The model is specifically trained for code generation and understanding across multiple programming languages.
+
+## Key Features
+- High-performance inference with vLLM
+- OpenAI-compatible API
+- Specialized in code generation
+- Multi-language support
+- Optimized for GPU acceleration
+
+## Configuration
+- Port: 8000
+- GPU: Required (24GB+ VRAM)
+- Model: agentica-org/DeepCoder-1.5B-Preview
+- Context Length: 8192 tokens
+
+## Usage
+You can interact with the model using the OpenAI API format:
+```python
+from openai import OpenAI
+
+
+openai_api_key = "EMPTY"
+openai_api_base = "https://fletcsut7rq2xsjcqw8vvfldgopf273zizxmbdtwqrjp.node.k8s.prd.nos.ci/v1" # my deployed nosara job
+
+client = OpenAI(
+    api_key=openai_api_key,
+    base_url=openai_api_base,
+)
+completion = client.completions.create(model="agentica-org/DeepCoder-1.5B-Preview",
+                                      prompt="Write me a simple binary search in python" , max_tokens=5000)
+
+print(completion.choices[0].text)
+```
+
+## License
+This model is available for research and commercial use. Please check the model card for the most up-to-date licensing information.
+
+## Nosana Run
+https://dashboard.nosana.com/jobs/89JkqUAY3y1MULRp7QLRovguDad169CgBkfiVPSBKvxt

templates/DeepCoder-1.5B/info.json

@@ -0,0 +1,12 @@
+{
+    "id": "deepcoder-1.5b",
+    "name": "DeepCoder 1.5B",
+    "description": "High-performance code generation model by Agentica Org with OpenAI-compatible API endpoint",
+    "category": [
+        "AI",
+        "Code Generation",
+        "Development"
+    ],
+    "icon": "https://cdn-avatars.huggingface.co/v1/production/uploads/652867b09903f7a1c9f7cbf1/s5AyAWwXFutyA_GY0Ct1e.jpeg"
+}
+

templates/DeepCoder-1.5B/job-definition.json

@@ -0,0 +1,27 @@
+{
+    "version": "0.1",
+    "type": "container",
+    "meta": {
+      "trigger": "dashboard",
+      "system_requirements": {
+        "required_vram": 8
+      }
+    },
+    "ops": [
+      {
+        "type": "container/run",
+        "id": "deepcoder-1.5b",
+        "args": {
+          "entrypoint": [],
+          "cmd": [
+            "/bin/sh",
+            "-c",
+            "python3 -m vllm.entrypoints.openai.api_server --model agentica-org/DeepCoder-1.5B-Preview --port 9000 --max-model-len 32000"
+          ],
+          "image": "docker.io/vllm/vllm-openai:v0.7.2",
+          "gpu": true,
+          "expose": 9000
+        }
+      }
+    ]
+  } 

templates/Shap-E Text-to-3D/README.md

@@ -0,0 +1,75 @@
+# Shap-E Text to 3D Generator
+
+## Description
+This template provides an easy way to generate 3D models from text descriptions using OpenAI's Shap-E model via Huggingface. The model transforms text prompts into animated 3D objects and exports them as GIF files that show the object from multiple angles.
+
+## Features
+- Convert text descriptions to 3D models
+- Uses OpenAI's Shap-E model from Huggingface
+- GPU-accelerated for optimal performance
+- Outputs animated GIFs showing the 3D object from different angles
+- Customizable guidance scale and inference steps
+
+## Technical Details
+- Model: `openai/shap-e`
+- Default guidance scale: 15.0
+- Default inference steps: 64
+- Output format: Animated GIF
+
+## Requirements
+- GPU resources for efficient processing
+
+## Usage
+1. Enter your desired text prompt in the PROMPT environment variable
+2. Run the job through the Nosana dashboard
+3. Download the generated animated GIF from the output directory
+
+## Example Prompts
+
+<table align="center">
+    <tbody>
+        <tr>
+            <td align="center">
+                <img src="shark_3d.gif" alt="A shark">
+            </td>
+            <td align="center">
+                <img src="elephant_3d.gif" alt="An elephant">
+            </td align="center">      
+        </tr>
+        <tr>
+            <td align="center">A shark</td>
+            <td align="center">An elephant </td>  
+        </tr>
+    </tbody>
+<table>
+
+## How It Works
+The job uses the Shap-E diffusion model from OpenAI, which has been trained to generate 3D shapes from text descriptions. The model creates a 3D representation that is then rendered from multiple angles to create an animated GIF.
+
+```python
+import torch
+from diffusers import ShapEPipeline
+from diffusers.utils import export_to_gif
+
+ckpt_id = "openai/shap-e"
+pipe = ShapEPipeline.from_pretrained(ckpt_id).to("cuda")
+
+guidance_scale = 15.0
+prompt = "a shark"
+images = pipe(
+    prompt,
+    guidance_scale=guidance_scale,
+    num_inference_steps=64,
+).images
+
+gif_path = export_to_gif(images[0], "shark_3d.gif")
+```
+
+## Links
+- [Shap-E Model on Huggingface](https://huggingface.co/openai/shap-e)
+
+- Details of  Successful Nosara Run.
+
+ ![Example Successful Nosana Job](https://gcdnb.pbrd.co/images/Gz4dueL9DmbM.png)
+
+

templates/Shap-E Text-to-3D/info.json

@@ -0,0 +1,7 @@
+{
+    "id": "shap-e-text-to-3d",
+    "name": "Shap-E Text to 3D Generator",
+    "description": "Generate 3D models from text prompts using OpenAI's Shap-E model from Huggingface",
+    "category": ["AI", "3D Generation", "Huggingface"],
+    "icon": "https://gcdnb.pbrd.co/images/fVCxCflg7jxU.png"
+  }

templates/Shap-E Text-to-3D/job-definition.json

@@ -0,0 +1,28 @@
+{
+  "ops": [
+    {
+      "id": "shap-E",
+      "args": {
+        "cmd": [
+          "/bin/sh",
+          "-c",
+          "apt-get update && apt-get install -y git && git clone https://github.com/eskayML/shap-E-Text-to-3D && cd shap-E-Text-to-3D && pip install -r requirements.txt && pip install huggingface-hub && mkdir openai && cd openai && huggingface-cli download openai/shap-e --local-dir . --local-dir-use-symlinks False && cd ../ && python app.py --host 0.0.0.0 --port 7860"
+        ],
+        "env": {},
+        "gpu": true,
+        "image": "pytorch/pytorch:2.0.0-cuda11.7-cudnn8-runtime",
+        "expose": 7860,
+        "entrypoint": []
+      },
+      "type": "container/run"
+    }
+  ],
+  "meta": {
+    "trigger": "dashboard",
+    "system_requirements": {
+      "required_vram": 16
+    }
+  },
+  "type": "container",
+  "version": "0.1"
+}