udpate , use stream mode all the time

src/ApplyView.tsx

@@ -1,4 +1,4 @@
-import { TFile, View, WorkspaceLeaf } from 'obsidian'
+import { View, WorkspaceLeaf } from 'obsidian'
 import { Root, createRoot } from 'react-dom/client'
 
 import ApplyViewRoot from './components/apply-view/ApplyViewRoot'
@@ -27,7 +27,7 @@ export class ApplyView extends View {
 	}
 
 	getDisplayText() {
-		return `Applying: ${this.state?.file?.name ?? ''}`
+		return `Applying: ${this.state?.file ?? ''}`
 	}
 
 	async setState(state: ApplyViewState) {

src/components/apply-view/ApplyViewRoot.tsx

@@ -128,7 +128,7 @@ export default function ApplyViewRoot({
         </div>
         <div className="view-header-title-container mod-at-start">
           <div className="view-header-title">
-            Applying: {state?.file?.name ?? ''}
+            Applying: {state?.file ?? ''}
           </div>
           <div className="view-actions">
             <button
@@ -157,8 +157,8 @@ export default function ApplyViewRoot({
             <div className="cm-scroller">
               <div className="cm-sizer">
                 <div className="infio-inline-title">
-                  {state?.file?.name
+                  {state?.file
-                    ? state.file.name.replace(/\.[^/.]+$/, '')
+                    ? state.file.replace(/\.[^/.]+$/, '')
                     : ''}
                 </div>
 

src/components/inline-edit/InlineEdit.tsx

@@ -217,21 +217,34 @@ export const InlineEdit: React.FC<InlineEditProps> = ({
 				endLine: defaultEndLine,
 			});
 
-			const response = await llmManager.generateResponse(chatModel, {
+			const stream = await llmManager.streamResponse(
-				model: chatModel.modelId,
+				chatModel,
-				messages: requestMessages,
+				{
-				stream: false,
+					messages: requestMessages,
-			});
+					model: chatModel.modelId,
+					max_tokens: settings.modelOptions.max_tokens,
+					temperature: settings.modelOptions.temperature,
+					// top_p: settings.modelOptions.top_p,
+					// frequency_penalty: settings.modelOptions.frequency_penalty,
+					// presence_penalty: settings.modelOptions.presence_penalty,
+					stream: true,
+				}
+			)
 
-			if (!response.choices[0].message.content) {
+			let response_content = ""
+			for await (const chunk of stream) {
+				const content = chunk.choices[0]?.delta?.content ?? ''
+				response_content += content
+			}
+			if (!response_content) {
 				setIsSubmitting(false);
 				throw new Error("Empty response from LLM");
 			}
 
 			const parsedBlock = parseSmartComposeBlock(
-				response.choices[0].message.content
+				response_content
 			);
-			const finalContent = parsedBlock?.content || response.choices[0].message.content;
+			const finalContent = parsedBlock?.content || response_content;
 
 			if (!activeFile || !(activeFile.path && typeof activeFile.path === 'string')) {
 				setIsSubmitting(false);

src/core/autocomplete/index.ts

@@ -33,12 +33,37 @@ class LLMClient {
 	}
 
 	async queryChatModel(messages: RequestMessage[]): Promise<Result<string, Error>> {
-		const data = await this.llm.generateResponse(this.model, {
+		const stream = await this.llm.streamResponse(
+			this.model,
+			{
+				messages: messages,
+				model: this.model.modelId,
+				stream: true,
+			}
+		)
+
+		let response_content = ""
+		for await (const chunk of stream) {
+			const content = chunk.choices[0]?.delta?.content ?? ''
+			response_content += content
+		}
+		return ok(response_content);
+	}
+
+	async queryChatModelStream(messages: RequestMessage[]): Promise<AsyncIterable<string>> {
+		const stream = await this.llm.streamResponse(this.model, {
 			model: this.model.modelId,
 			messages: messages,
-			stream: false,
+			stream: true,
 		})
-		return ok(data.choices[0].message.content);
+		// eslint-disable-next-line no-inner-declarations
+		async function* streamResponse(): AsyncIterable<string> {
+			for await (const chunk of stream) {
+				yield chunk.choices[0].delta.content;
+			}
+		}
+
+		return streamResponse()
 	}
 }