Update app.py

app.py

@@ -1,7 +1,6 @@
 import discord
 from discord.ext import commands
 from ChatAI.chat_ai import pipe as ai
-from huggingface_hub import InferenceClient
 
 
 # Set up Discord bot intents and command prefix
@@ -12,7 +11,6 @@ bot = commands.Bot(command_prefix="!", intents=intents)
 # Dictionary to track message count per channel
 message_counts = {}
 
-
 @bot.event
 async def on_message(message):
     guild = message.guild  # Get the guild (server) the message is from
@@ -30,30 +28,43 @@ async def on_message(message):
     print(message_counts[message.channel.id])
 
     messages = []
-    if message_counts[message.channel.id] >= 10: # Check if the count reaches 10
-        async for message in channel.history(limit=10):
+    if message_counts[message.channel.id] >= 1: # Check if the count reaches 10
+        async for message in channel.history(limit=4):
             messages.append(message.content)
-        previous_messages = ("\n".join(messages))
 
-        response = ai.text_generation(previous_messages, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0)
-        splitted_response = split_string(response)
-        for part in splitted_response:
+        previous_messages = ("\n".join(messages))
+        response = generate(previous_messages, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0)
+        part = split_string(str(response))
+        for part in part:
             await channel.send(part)
 
         message_counts[message.channel.id] = 0 # Reset the counter
 
-
-    response = str(ai.text_generation(message, details=False, stream=False, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0))
-    splitted_response = split_string(response)
-    for part in splitted_response:
-        await channel.send(part)
-
     await bot.process_commands(message) # Ensure commands still work
 
 def split_string(text: str) -> list[str]:
     """Helper function to split text into chunks"""
     return [text[i:i+3900] for i in range(0, len(text), 3900)]
 
+def generate(
+    prompt, temperature=0.9, max_new_tokens=256, top_p=0.95, repetition_penalty=1.0
+):
+    temperature = float(temperature)
+    if temperature < 1e-2:
+        temperature = 1e-2
+    top_p = float(top_p)
+
+    generate_kwargs = dict(
+        temperature=temperature,
+        max_new_tokens=max_new_tokens,
+        top_p=top_p,
+        repetition_penalty=repetition_penalty,
+        do_sample=True,
+        seed=42,
+    )
+
+    return ai.text_generation(prompt, **generate_kwargs, stream=False, details=False, return_full_text=True)
+
 @bot.event
 async def on_ready():
     print(f'Logged in as {bot.user}')  # Logs bot login in console