Reorg for CPU hosting

Makefile

@@ -0,0 +1,5 @@
+start-gpu:
+	docker compose -f docker-compose.yml -f docker-compose-gpu.yml up
+
+start-cpu:
+	docker compose -f docker-compose.yml -f docker-compose-cpu.yml up

docker-compose-cpu.yml

@@ -0,0 +1,8 @@
+version: '3.8'
+
+services:
+  serving:
+    deploy:
+      resources:
+        reservations:
+          devices: []

docker-compose-gpu.yml

@@ -0,0 +1,11 @@
+version: '3.8'
+
+services:
+  serving:
+    deploy:
+     resources:
+       reservations:
+         devices:
+           - driver: nvidia
+             device_ids: ["${GPU_DEVICE}"]
+             capabilities: [ gpu ]

docker-compose.yml

@@ -9,13 +9,8 @@ services:
     userns_mode: "host"
     volumes:
       - ./models:/models:z
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              device_ids: ["${GPU_DEVICE}"]
-              capabilities: [ gpu ]
+    ports:
+      - "8501:8501"
     shm_size: 8GB
 
   develop:

models/3dmolms/3dmolms_test/1/model.py

@@ -0,0 +1,25 @@
+import json
+import triton_python_backend_utils as pb_utils
+
+
+class TritonPythonModel:
+    def __init__(self):
+        super().__init__()
+        self.output_dtype = []
+
+    def initialize(self, args):
+        model_config = json.loads(args["model_config"])
+        output0_config = pb_utils.get_output_config_by_name(model_config, "charge_norm")
+        self.output_dtype = pb_utils.triton_string_to_numpy(output0_config["data_type"])
+
+    def execute(self, requests):
+        responses = []
+        for request in requests:
+            raw = pb_utils.get_input_tensor_by_name(request, "charge_raw")
+            norm = raw.as_numpy() * 0.1
+            ce_tensor = pb_utils.Tensor("charge_norm", norm.astype(self.output_dtype))
+            responses.append(pb_utils.InferenceResponse(output_tensors=[ce_tensor]))
+        return responses
+
+    def finalize(self):
+        pass

models/3dmolms/3dmolms_test/config.pbtxt

@@ -0,0 +1,15 @@
+max_batch_size: 1000
+input[
+ {
+   name: 'charge_raw',
+   data_type: TYPE_INT32,
+   dims: [1]
+ }
+]
+output [
+ {
+   name: 'charge_norm',
+   data_type: TYPE_FP32,
+   dims: [1]
+ }
+]