VAREK
————

-- Schema, pipeline, and logic in one file
schema ImageInput { 
  path:   str, 
  label:  str?, 
  width:  int, 
  height: int 
} 

pipeline classify_images { 
  source: ImageInput[] 
  steps:  [preprocess -> embed -> infer] 
  output: ClassResult[] 
} 

fn preprocess(img: ImageInput) -> Tensor { 
  load_image(img.path) 
    |> resize(224, 224) 
    |> normalize(mean=[0.485, 0.456]) 
} 

async fn infer(t: Tensor) -> ClassResult { 
  let model = load_model("resnet50.synmodel") 
  model.forward(t) 
}

# schema.json + config.yaml + pipeline.py + run.sh

# --- schema.json ---
{ 
  "type": "object", 
  "properties": { 
    "path": {"type": "string"}, 
    "label": {"type": "string"} 
  }
} 

# --- pipeline.py ---
from torchvision import transforms 
from PIL import Image 
import torch, yaml, json, subprocess 

def preprocess(img_path): 
  t = transforms.Compose([ 
    transforms.Resize((224, 224)), 
    transforms.ToTensor(), 
    transforms.Normalize([0.485, 0.456]) 
  ]) 
  return t(Image.open(img_path)) 

async def infer(tensor): 
  model = torch.load("resnet50.pt") 
  model.eval() 
  with torch.no_grad(): 
    return model(tensor.unsqueeze(0))