Updating templates

templates/components/llamafile.jsonnet

@@ -6,12 +6,14 @@ local prompts = import "prompts/slm.jsonnet";
 {
 
     "llamafile-model":: "LLaMA_CPP",
-    "llamafile-url":: "${LLAMAFILE_URL}",
 
     "text-completion" +: {
     
         create:: function(engine)
 
+            local envSecrets = engine.envSecrets("llamafile-credentials")
+                .with_env_var("LLAMAFILE_URL", "llamafile-url");
+
             local container =
                 engine.container("text-completion")
                     .with_image(images.trustgraph)
@@ -21,27 +23,12 @@ local prompts = import "prompts/slm.jsonnet";
                         url.pulsar,
                         "-m",
                         $["llamafile-model"],
-                        "-r",
-                        $["llamafile-url"],
                     ])
+                    .with_env_var_secrets(envSecrets)
                     .with_limits("0.5", "128M")
                     .with_reservations("0.1", "128M");
 
-            local containerSet = engine.containers(
-                "text-completion", [ container ]
-            );
-
-            engine.resources([
-                containerSet,
-            ])
-
-    },
-
-    "text-completion-rag" +: {
-    
-        create:: function(engine)
-
-            local container =
+            local containerRag =
                 engine.container("text-completion-rag")
                     .with_image(images.trustgraph)
                     .with_command([
@@ -50,26 +37,40 @@ local prompts = import "prompts/slm.jsonnet";
                         url.pulsar,
                         "-m",
                         $["llamafile-model"],
-                        "-r",
-                        $["llamafile-url"],
                         "-i",
                         "non-persistent://tg/request/text-completion-rag",
                         "-o",
                         "non-persistent://tg/response/text-completion-rag-response",
                     ])
+                    .with_env_var_secrets(envSecrets)
                     .with_limits("0.5", "128M")
                     .with_reservations("0.1", "128M");
 
             local containerSet = engine.containers(
-                "text-completion-rag", [ container ]
+                "text-completion", [ container ]
             );
 
+            local containerSetRag = engine.containers(
+                "text-completion-rag", [ containerRag ]
+            );
+
+            local service =
+                engine.internalService(containerSet)
+                .with_port(8080, 8080, "metrics");
+
+            local serviceRag =
+                engine.internalService(containerSetRag)
+                .with_port(8080, 8080, "metrics");
+
             engine.resources([
+                envSecrets,
                 containerSet,
+                containerSetRag,
+                service,
+                serviceRag,
             ])
 
-
-    }
+    },
 
 } + prompts
 

trustgraph-bedrock/trustgraph/model/text_completion/bedrock/llm.py

@@ -22,7 +22,7 @@ default_input_queue = text_completion_request_queue
 default_output_queue = text_completion_response_queue
 default_subscriber = module
 default_model = 'mistral.mistral-large-2407-v1:0'
-default_region = 'us-west-2'
+default_region = os.getenv("AWS_REGION", 'us-west-2')
 default_temperature = 0.0
 default_max_output = 2048
 default_aws_id = os.getenv("AWS_ID_KEY")
@@ -307,7 +307,7 @@ class Processor(ConsumerProducer):
 
         parser.add_argument(
             '-r', '--aws-region',
-            help=f'AWS Region (default: us-west-2)'
+            help=f'AWS Region'
         )
 
         parser.add_argument(
@@ -328,4 +328,3 @@ def run():
 
     Processor.start(module, __doc__)
 
-    

trustgraph-flow/trustgraph/model/text_completion/llamafile/llm.py

@@ -20,7 +20,7 @@ default_input_queue = text_completion_request_queue
 default_output_queue = text_completion_response_queue
 default_subscriber = module
 default_model = 'LLaMA_CPP'
-default_llamafile = 'http://localhost:8080/v1'
+default_llamafile = os.getenv("LLAMAFILE_URL", "http://localhost:8080/v1")
 default_temperature = 0.0
 default_max_output = 4096