Update app.py

app.py

@@ -15,13 +15,13 @@ try:
     tokenizer = PreTrainedTokenizerFast.from_pretrained(MODEL_ID)
     
     # Ensure the tokenizer has the necessary special tokens
-    #special_tokens = {
+    special_tokens = {
     #    'pad_token': '[PAD]',
-    #    'eos_token': '</s>',
+        'eos_token': '<|eot_id|>'
     #    'bos_token': '<s>',
     # 3   'unk_token': '<unk>'
-    #}
-    #tokenizer.add_special_tokens(special_tokens)
+    }
+    tokenizer.add_special_tokens(special_tokens)
     
     model = AutoModelForCausalLM.from_pretrained(MODEL_ID, gguf_file=GGUF_FILE)
 
@@ -47,11 +47,19 @@ def format_diff_response(response):
     return '<br>'.join(formatted)
 
 def create_prompt(request, file_content, system_message):
-    return f"""<system>{system_message}</system>
+#    return f"""<system>{system_message}</system>
+#<request>{request}</request>
+#<file>{file_content}</file>"""
+    return f"""<|begin_of_text|><|start_header_id|>system<|end_header_id|>
+
+Cutting Knowledge Date: December 2023
+Today Date: 26 July 2024
+
+{system_message}<|eot_id|><|start_header_id|>user<|end_header_id|>
+
 <request>{request}</request>
-<file>
-{file_content}
-</file>"""
+#<file>{file_content}</file><|eot_id|><|start_header_id|>assistant<|end_header_id|>
+"""
 
 @spaces.GPU
 def respond(request, file_content, system_message, max_tokens, temperature, top_p):
@@ -122,7 +130,7 @@ with gr.Blocks() as demo:
         max_tokens = gr.Slider(
             minimum=1,
             maximum=2048,
-            value=512,
+            value=128,
             step=1,
             label="Max Tokens"
         )
@@ -130,13 +138,13 @@ with gr.Blocks() as demo:
             minimum=0.1,
             maximum=4.0,
             value=0.7,
-            step=0.1,
+            step=1,
             label="Temperature"
         )
         top_p = gr.Slider(
             minimum=0.1,
             maximum=1.0,
-            value=0.95,
+            value=1,
             step=0.05,
             label="Top-p"
         )