SearchGPT: Enhance. #3

config.py

@@ -8,6 +8,8 @@
 
 MODEL = "gpt-4.1-nano"
 
+MAX_TOKENS = 131072
+
 SEARXNG_ENDPOINT = "https://searx.stream/search"  # See the endpoint list at https://searx.space
 BAIDU_ENDPOINT   = "https://www.baidu.com/s"
 READER_ENDPOINT  = "https://r.jina.ai/"
@@ -240,7 +242,7 @@ REASONING_STEPS = {
 
 REASONING_DEFAULT = "I'm processing the tool execution request..."
 
-REASONING_DELAY = 0.15  # 150 ms
+REASONING_DELAY = 0.01  # 10 ms
 
 OS = [
     "Windows NT 10.0; Win64; x64",

src/processor/reasoning/interface.py

@@ -5,5 +5,14 @@
 
 def reasoning_interfaces(text, current_length=0):
     if current_length < len(text):
-        return text[:current_length]
+        reasoning_steps = text[:current_length]
+        
+        if current_length > 0 and not reasoning_steps.endswith((
+            '<br>',
+            '<br><br>'
+        )):
+            reasoning_steps += '...'
+        
+        return reasoning_steps
+    
     return text

src/processor/response/generator.py

@@ -4,7 +4,7 @@
 #
 
 import traceback
-from config import INSTRUCTIONS_END
+from config import MAX_TOKENS, INSTRUCTIONS_END
 
 def generate_response(
     server,
@@ -28,6 +28,7 @@ def generate_response(
             messages=conversation_messages,
             tools=tool_definitions if not tools_done else None,
             tool_choice="none",
+            max_tokens=MAX_TOKENS,
             temperature=0.75,
             stream=True
         )

src/processor/tools/interaction.py

@@ -10,7 +10,7 @@ from ..reasoning.interface import reasoning_interfaces
 from ..reasoning.tool_reasoning import tool_reasoning
 from .parser import extract_tool_parameters
 from .executor import invoke_tool_function
-from config import REASONING_DELAY
+from config import MAX_TOKENS, REASONING_DELAY
 
 def process_tool_interactions(server, model_name, conversation_messages, tool_definitions, search_engine):
     maximum_iterations = 1
@@ -24,6 +24,7 @@ def process_tool_interactions(server, model_name, conversation_messages, tool_de
                 messages=conversation_messages,
                 tools=tool_definitions,
                 tool_choice="auto",
+                max_tokens=MAX_TOKENS,
                 temperature=0.6
             )
         except Exception:
@@ -55,7 +56,7 @@ def process_tool_interactions(server, model_name, conversation_messages, tool_de
             
             if extraction_error:
                 error_reasoning = tool_reasoning(tool_name, None, "error", error=extraction_error)
-                for i in range(0, len(error_reasoning), 50):
+                for i in range(0, len(error_reasoning), 5):
                     logs_generator = styles(reasoning_interfaces(error_reasoning, i), expanded=True)
                     yield logs_generator
                     time.sleep(REASONING_DELAY)
@@ -64,13 +65,13 @@ def process_tool_interactions(server, model_name, conversation_messages, tool_de
                 tool_execution_result = extraction_error
             else:
                 reasoning_status = tool_reasoning(tool_name, extracted_arguments, "parsing")
-                for i in range(0, len(reasoning_status), 50):
+                for i in range(0, len(reasoning_status), 5):
                     logs_generator = styles(reasoning_interfaces(reasoning_status, i), expanded=True)
                     yield logs_generator
                     time.sleep(REASONING_DELAY)
                 
                 reasoning_start = tool_reasoning(tool_name, extracted_arguments, "executing")
-                for i in range(0, len(reasoning_start), 50):
+                for i in range(0, len(reasoning_start), 5):
                     logs_generator = styles(reasoning_interfaces(reasoning_start, i), expanded=True)
                     yield logs_generator
                     time.sleep(REASONING_DELAY)
@@ -88,7 +89,7 @@ def process_tool_interactions(server, model_name, conversation_messages, tool_de
                     })
                     
                     reasoning_done = tool_reasoning(tool_name, extracted_arguments, "completed", result=tool_execution_result)
-                    for i in range(0, len(reasoning_done), 50):
+                    for i in range(0, len(reasoning_done), 5):
                         logs_generator = styles(reasoning_interfaces(reasoning_done, i), expanded=True)
                         yield logs_generator
                         time.sleep(REASONING_DELAY)
@@ -97,7 +98,7 @@ def process_tool_interactions(server, model_name, conversation_messages, tool_de
                     
                 except Exception as tool_error:
                     error_reasoning = tool_reasoning(tool_name, extracted_arguments, "error", error=str(tool_error))
-                    for i in range(0, len(error_reasoning), 50):
+                    for i in range(0, len(error_reasoning), 5):
                         logs_generator = styles(reasoning_interfaces(error_reasoning, i), expanded=True)
                         yield logs_generator
                         time.sleep(REASONING_DELAY)