s

app.py

@@ -34,8 +34,8 @@ def sanitize_input(text):
     return html.escape(text)
 
 def validate_parameters(max_tokens, temperature, top_p):
-    if not (1 <= max_tokens <= 2048):
-        return False, "Error: 'Max new tokens' must be between 1 and 2048."
+    if not (1 <= max_tokens <= 1024):
+        return False, "Error: 'Max new tokens' must be between 1 and 1024."
     if not (0.1 <= temperature <= 4.0):
         return False, "Error: 'Temperature' must be between 0.1 and 4.0."
     if not (0.1 <= top_p <= 1.0):
@@ -43,15 +43,12 @@ def validate_parameters(max_tokens, temperature, top_p):
     return True, ""
 
 # Load the model and tokenizer
-model_name = "HuggingFaceH4/mistral-7b-instruct"  # Update with the correct model name
+model_name = "gpt2"  # Use GPT-2 model
 
 try:
-    from transformers import MistralForCausalLM, MistralTokenizer
-
-    tokenizer = MistralTokenizer.from_pretrained(model_name)
-    model = MistralForCausalLM.from_pretrained(
+    tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
+    model = transformers.AutoModelForCausalLM.from_pretrained(
         model_name,
-        torch_dtype=torch.float16,
         device_map="auto",
     )
     model.eval()
@@ -69,6 +66,7 @@ def respond(message, history, persona_choice, custom_persona, max_tokens, temper
     truncated_history = safe_history[-MAX_HISTORY_LENGTH:]
     system_message = system_message_selector(persona_choice, custom_persona)
 
+    # Build the conversation prompt
     conversation = system_message + "\n\n"
     for user_msg, bot_msg in truncated_history:
         conversation += f"User: {user_msg}\n"
@@ -90,7 +88,9 @@ def respond(message, history, persona_choice, custom_persona, max_tokens, temper
             eos_token_id=tokenizer.eos_token_id,
         )
 
-        generated_text = tokenizer.decode(output_ids[0][input_ids.shape[-1]:], skip_special_tokens=True)
+        generated_text = tokenizer.decode(
+            output_ids[0][input_ids.shape[-1]:], skip_special_tokens=True
+        )
         return generated_text.strip()
     except Exception as e:
         logging.error(f"An error occurred: {e}")
@@ -109,15 +109,15 @@ system_message_textbox = gr.Textbox(
 )
 
 max_tokens_slider = gr.Slider(
-    minimum=1, maximum=1024, value=512, step=1, label="Max new tokens"
+    minimum=1, maximum=1024, value=50, step=1, label="Max new tokens"
 )
 
 temperature_slider = gr.Slider(
-    minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"
+    minimum=0.1, maximum=4.0, value=1.0, step=0.1, label="Temperature"
 )
 
 top_p_slider = gr.Slider(
-    minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"
+    minimum=0.1, maximum=1.0, value=0.9, step=0.05, label="Top-p (nucleus sampling)"
 )
 
 # Create the ChatInterface

requirements.txt

@@ -1,4 +1,3 @@
-transformers>=4.34.0
+transformers==4.31.0
 gradio==3.40.1
-torch>=2.0.1
-xformers
+torch==2.0.1