Update app.py

app.py

@@ -8,6 +8,7 @@ import logging
 from unsloth import FastLanguageModel
 import subprocess
 
+# Set up logging for debugging
 logging.basicConfig(
     level=logging.DEBUG,  # Set the logging level to DEBUG to capture all messages
     format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
@@ -17,7 +18,11 @@ logging.basicConfig(
 )
 logger = logging.getLogger(__name__)
 logger.info("HELLO WORLD...")
+
+# Get environment variable for Hugging Face access
 READ_HF = os.environ["read_hf"]
+
+# Alpaca prompt template
 alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
 
 ### Instruction:
@@ -28,6 +33,8 @@ alpaca_prompt = """Below is an instruction that describes a task, paired with an
 
 ### Response:
 {}"""
+
+# Inventory management instructions
 string = '''
 You are an AI assistant tasked with managing inventory based on user instructions. You must meticulously analyze each user request to determine the appropriate action and execute it with the correct parameters.
 
@@ -72,6 +79,7 @@ You are an AI assistant tasked with managing inventory based on user instruction
 
 Category List : ["Dairy & Eggs", "Beverages & Snacks", "Cleaning & Hygiene", "Grains & Staples", "Personal Care", "Other"]
 '''
+
 @spaces.GPU()
 def chunk_it(inventory_list, user_input_text):
     # Check for CUDA and NVIDIA-related errors
@@ -98,46 +106,42 @@ def chunk_it(inventory_list, user_input_text):
             token = READ_HF    
         )
         logger.info("Model and tokenizer loaded.")
+
+        # ... (rest of your code)
+
+        formatted_prompt = alpaca_prompt.format(
+            string + inventory_list,  # instruction
+            user_input_text,  # input
+            "",  # output - leave this blank for generation!
+        )
+        logger.debug(f"Formatted prompt: {formatted_prompt}")
+        try:
+            inputs = tokenizer([formatted_prompt], return_tensors="pt").to("cuda")
+            logger.debug(f"Tokenized inputs: {inputs}")
+        except Exception as e:
+            logger.error(f"Failed to tokenize inputs: {e}")
+            raise
+        logger.info("Generating output...")
+        try:
+            outputs = model.generate(**inputs, max_new_tokens=216, use_cache=True)
+            logger.info("Output generated.")
+        except Exception as e:
+            logger.error(f"Failed to generate output: {e}")
+            raise
+        try:
+            reply = tokenizer.batch_decode(outputs, skip_special_tokens=True)
+            logger.debug(f"Decoded output: {reply}")
+        except Exception as e:
+            logger.error(f"Failed to decode output: {e}")
+            raise
+        logger.debug(f"Final reply: {reply}")
+        return reply
+
     except Exception as e:
-        logger.error(f"Failed to load model and tokenizer: {e}")
-        raise
-    logger.info("Enabling native 2x faster inference...")
-    try:
-        FastLanguageModel.for_inference(model)
-        logger.info("Inference enabled.")
-    except Exception as e:
-        logger.error(f"Failed to enable native inference: {e}")
-        raise
-    formatted_prompt = alpaca_prompt.format(
-        string + inventory_list,  # instruction
-        user_input_text,  # input
-        "",  # output - leave this blank for generation!
-    )
-    logger.debug(f"Formatted prompt: {formatted_prompt}")
-    try:
-        inputs = tokenizer([formatted_prompt], return_tensors="pt").to("cuda")
-        logger.debug(f"Tokenized inputs: {inputs}")
-    except Exception as e:
-        logger.error(f"Failed to tokenize inputs: {e}")
-        raise
-    logger.info("Generating output...")
-    try:
-        outputs = model.generate(**inputs, max_new_tokens=216, use_cache=True)
-        logger.info("Output generated.")
-    except Exception as e:
-        logger.error(f"Failed to generate output: {e}")
-        raise
-    try:
-        reply = tokenizer.batch_decode(outputs, skip_special_tokens=True)
-        logger.debug(f"Decoded output: {reply}")
-    except Exception as e:
-        logger.error(f"Failed to decode output: {e}")
-        raise
-    # pattern = r"### Response:\n(.*?)<\|end_of_text\|>"
-    # match = re.search(pattern, reply[0], re.DOTALL)
-    # reply = match.group(1).strip()
-    logger.debug(f"Final reply: {reply}")
-    return reply
+        logger.error(f"Error loading model or CUDA issues: {e}")
+        return "There seems to be an issue with CUDA or the model. Please check the Hugging Face Spaces environment."
+
+
 # Interface for inputs
 iface = gr.Interface(
     fn=chunk_it,
@@ -148,6 +152,15 @@ iface = gr.Interface(
     outputs=gr.Textbox(label="output", lines=23),
     title="Testing",
 )
+
+# Set up logging to display in Gradio
+logger = logging.getLogger(__name__)
+logger.setLevel(logging.DEBUG)  # Set the logging level
+ch = logging.StreamHandler(gr.Log())  # Create a StreamHandler and send logs to gr.Log
+formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
+ch.setFormatter(formatter)
+logger.addHandler(ch)
+
 logger.info("Launching Gradio interface...")
 try:
     iface.launch(inline=False)