update

Assistant.py

@@ -11,7 +11,7 @@ class Assistant:
     def __init__(self, default_model_path, default_chat_format):
         self.model = None
         self.max_context_tokens = 2048
-        self.max_assistant_tokens = 2048
+        self.max_assistant_tokens = 1024
         self.is_processing = False
         self.chat_format = default_chat_format
         self.model_path = default_model_path
@@ -50,19 +50,19 @@ def load_default_model(self):
         self.context_window_start = 0
         self.stop_emit = False
 
-    def load_model(self, model_path, format, new_temperature, n_gpu_layer, new_system_message, context):
+    def load_model(self, model_path, format, new_temperature, n_gpu_layer, new_system_message, context,max_response_tokens):
 
 
         message = new_system_message if isinstance(new_system_message, str) and new_system_message != '' else self.default_system_message
         gpu_layers = int(n_gpu_layer) if isinstance(n_gpu_layer, int) else self.gpu_layers
         ctx = context if isinstance(context, int)  else self.max_context_tokens
         temperature = new_temperature if isinstance(new_temperature, float) else self.temperature
         self.default_system_message = new_system_message
-        print(new_system_message)
+        max_asistant_tokens = max_response_tokens if isinstance(max_response_tokens,int) else self.max_assistant_tokens
         self.model_path = model_path
         self.temperature = temperature
         self.max_context_tokens = ctx
-        self.max_assistant_tokens = self.max_assistant_tokens #TODO: change in interface
+        self.max_assistant_tokens = max_asistant_tokens #TODO: change in interface
         self.chat_format = format
         self.gpu_layers = gpu_layers
         self.stop_emit = False

FlaskApi.py

@@ -124,16 +124,18 @@ def load_model(self):
         n_gpu_layers = int(request.form.get('gpu_layers')) if request.form.get('gpu_layers') != '' else -1
         system_message = request.form.get('system_message')
         temperature = float(request.form.get('temperature')) if request.form.get('temperature') != '' else 0.81
+        max_response_tokens = request.form.get('max_response_tokens')
         n_ctx = int(request.form.get('context')) if request.form.get('context') != '' else 2048
         self.assistant.unload_model()
-        self.assistant.load_model(selected_model,selected_format,temperature,n_gpu_layers,system_message,n_ctx)
+        self.assistant.load_model(selected_model,selected_format,temperature,n_gpu_layers,system_message,n_ctx,max_response_tokens)
         return f'''
                 \nModel:{selected_model}
                 \nformat: {selected_format}
                 \ntemp: {temperature}
                 \nlayers: {n_gpu_layers}
                 \nSM: {system_message}
                 \nctx: {n_ctx}
+                \nmax tokens: {n_ctx}
                 '''
 
     def unload_model(self):

static/js/chat.js

@@ -15,7 +15,7 @@ class Chat {
         this.totalTokensResponse =0;
         this.conversationStarted = false;
         this.chatId = ' ';
-        this.library = 'ollama';
+        this.library = 'llama';
         this.adjustTextareaHeight();
         textarea.addEventListener('input', () => this.adjustTextareaHeight());  
         textarea.addEventListener('keydown', (e) => {
@@ -51,7 +51,7 @@ class Chat {
             //console.log("Tokens Respuesta: ", totalAssistantTokens);
             console.log(response)
         }else{
-            //const responseData = response.content["choices"][0];
+            const responseData = response.content["choices"][0];
             const { id, model, created, object } = response.content;
             const { index, delta, finish_reason } = responseData;
             const responseId = id;
@@ -266,7 +266,6 @@ class Chat {
 
     sendMessage() {
         if (!this.conversationStarted){
-
             this.conversationStarted= true;
             this.currentResponse = ' ';
             this.n_responses += 1;
@@ -298,23 +297,17 @@ class Chat {
                     var conversationListDiv = $('#conversations-list');
                     var buttonExists = false;
                     $('.load-history').each(function() {
-                        console.log($(this).text())
                         if ($(this).text() === '❌'+self.chatId) {
-
                             buttonExists = true;
-                            return false; // Salir del bucle each() si se encuentra un botón con el mismo texto
+                            return false; 
                         }
                     });
-                    console.log("Valor de self.chatId:", self.chatId);
-                    console.log("Número de botones existentes:", $('.load-history').length);
                     var conversationListDiv = $('#conversations-list');
                     var newChatHistory ='';
-
                     if(!buttonExists) {
                       newChatHistory = $("<div class='load-history' id='"+self.chatId+"'><button height='1em' width='1em' onclick=chat.deleteHistory('"+self.chatId+"')>❌</button><button  onclick=chat.loadHistory('"+self.chatId+"')>"+self.chatId+"</button></div>"); // $("<button class='load-history' onclick=chat.loadHistory('"+self.chatId+"')>📪 "+self.chatId+"</button>") 
                       conversationListDiv.prepend(newChatHistory);
-                    }
-
+                    }   
                     self.guardarHistorial(self.chatId , self.conversationHistory);
                     self.showPopup(data);
                     console.log(data);
@@ -366,6 +359,7 @@ class Chat {
         var gpuLayers = $('#gpu-layers').val();
         var temperature = $('#temperature').val();
         var n_ctx = $('#context').val();
+        var max_response_tokens = $('max-response-tokens').val();
         this.systemMessage = systemMessage;
         const self = this;
         $.ajax({
@@ -378,7 +372,8 @@ class Chat {
                 temperature: temperature,
                 system_message: systemMessage,
                 gpu_layers: gpuLayers,
-                context: n_ctx
+                context: n_ctx,
+                max_response_tokens : max_response_tokens
             },
 
             success: function (data) {

templates/index.html

@@ -79,6 +79,10 @@ <h5>Model configuration</h5>
             <label for="context">Max. context:</label>
             <input type="number" id="context" class="form-control" placeholder="Example: 2048">
         </div>
+        <div class="form-group">
+            <labell for="max-response-tokens">Max. Response Tokens</labell>
+            <input type="number" id="max-response-tokens" class="form-control" placeholder="Example: 1024">
+        </div>
         <div class="form-group">
             <label for="system-message">System message:</label>
             <textarea class="form-control" id="system-message" rows="3" placeholder="Example: You are a assistant with a friendly and honest personality. As an expert programmer and pentester, you should examine the details provided to ensure that they are usable.