nit: rename falcon

pkg/presets/falcon/inference-api.py

@@ -14,10 +14,10 @@
 import torch
 import torch.distributed as dist
 
-parser = argparse.ArgumentParser(description='HuggingFace Model Configuration')
+parser = argparse.ArgumentParser(description='Falcon Model Configuration')
 parser.add_argument('--load_in_8bit', default=False, action='store_true', help='Load model in 8-bit mode')
 parser.add_argument('--disable_trust_remote_code', default=False, action='store_true', help='Disable trusting remote code when loading the model')
-parser.add_argument('--model_id', required=True, type=str, help='The HuggingFace ID for the pre-trained model')
+parser.add_argument('--model_id', required=True, type=str, help='The Falcon ID for the pre-trained model')
 args = parser.parse_args()
 
 app = FastAPI()

pkg/presets/llama-webserver/pod.yaml

@@ -0,0 +1,36 @@
+# deployment.yaml
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: file-download-app
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: file-download-app
+  template:
+    metadata:
+      labels:
+        app: file-download-app
+    spec:
+      containers:
+      - name: my-container
+        image: smritidahal/file-download-app:latest
+        ports:
+        - containerPort: 80
+---
+# service.yaml
+apiVersion: v1
+kind: Service
+metadata:
+  name: file-download-service
+spec:
+  selector:
+    app: file-download-app
+  ports:
+    - protocol: TCP
+      port: 80
+      targetPort: 5000
+  type: LoadBalancer
+
+