more config

TODO.md

@@ -3,4 +3,10 @@
 [] - add fill in middle
 [x] - add config option to disable the extension
 [] - add command to test and query connection to server
-[x] - add feedback when waiting response
+[x] - add feedback when waiting response
+[] - add a chat window
+[] - if the model is an instruct-type add the system prompt to the chat
+[] - add an icon
+[] - option to backup and restore model settings
+[] - add a window to quickly modify model configs
+[] - decorate ai generated text https://github.com/microsoft/vscode-extension-samples/tree/main/decorator-sample

package.json

@@ -155,9 +155,18 @@
                 "dumbpilot.llamaTailfree_z": {"type": "number", "default": 0.5}, 
                 "dumbpilot.llamaSeed": {"type": "number", "default": -1},
                 "dumbpilot.llamaCachePrompt": {
-                    "type": "bool",
+                    "type": "boolean",
                     "default": true,
                     "description": "Enable prompt caching for faster results"
+                },
+                "dumbpilot.llamaInstructModel": {
+                    "type": "boolean",
+                    "default": "false",
+                    "description": "For use with instruct models"
+                },
+                "dumbpilot.llamaSystemPrompt": {
+                    "type": "string",
+                    "description": "The system prompt that the model considers at the beginning of every request, used by instruct models"
                 }
             }
         }

src/extension.ts

@@ -111,17 +111,13 @@ export function activate(context: vscode.ExtensionContext) {
 		config = vscode.workspace.getConfiguration("dumbpilot");
 	}));
 
-	let completion_enabled: boolean = config.get("completionEnabled") as boolean;
-
 	// TODO: work with local configurations
 	let disposable = vscode.commands.registerCommand("dumbpilot.enableCompletion", () => {
-		completion_enabled = true;
 		config.update("completionEnabled", true);
 	});
 	context.subscriptions.push(disposable);
 
 	disposable = vscode.commands.registerCommand("dumbpilot.disableCompletion", () => {
-		completion_enabled = false;
 		config.update("completionEnabled", false);
 	});
 
@@ -131,11 +127,12 @@ export function activate(context: vscode.ExtensionContext) {
 	const provider: vscode.InlineCompletionItemProvider = {
 		async provideInlineCompletionItems(document, position, context, token) {
 
-			if (completion_enabled === false) {
+			// disable if predictive completion is disabled
+			if (config.get("completionEnabled") as boolean === false) {
 				return null;
 			}
 			
-			// Since for every completion we want to query the server, we want to filter out
+			// Since for every completion we will query the server, we want to filter out
 			// automatic completion invokes
 			if (context.triggerKind === vscode.InlineCompletionTriggerKind.Automatic) {
 				return null;
@@ -224,6 +221,9 @@ export function activate(context: vscode.ExtensionContext) {
 				}
 		
 				data = await response.json() as llamaData;
+				const gen_tokens = data.timings.predicted_n;
+				const gen_time = (data.timings.predicted_ms / 1000).toFixed(2);
+				showMessageWithTimeout(`predicted ${gen_tokens} tokens in ${gen_time} seconds`, 1500);
 
 			} catch (e: any) {
 				const err = e as TypeError;