Update app.py

app.py

@@ -1,29 +1,24 @@
-from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+from transformers import AutoModelForCausalLM, AutoTokenizer
 import gradio as gr
 import torch
 
-# Model name
-model_name = "deepseek-ai/deepseek-coder-6.7b-instruct"  
+# Load a smaller model that fits within 16GB RAM
+model_name = "deepseek-ai/deepseek-coder-1.3b-instruct"
 
-# Use quantization (4-bit) to reduce memory usage
-bnb_config = BitsAndBytesConfig(
-    load_in_4bit=True,  # Use 4-bit quantization
-    bnb_4bit_compute_dtype=torch.float16,  # Reduce precision
-    bnb_4bit_use_double_quant=True,  # Further optimize memory
-)
-
-# Load model with optimizations
+# Load tokenizer
 tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+# Load model in CPU-friendly format (low precision for efficiency)
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
-    quantization_config=bnb_config,
-    device_map="auto"  # Automatically chooses best device (CPU/GPU)
+    torch_dtype=torch.float32,  # Use float32 since CPU-only
+    device_map="cpu"  # Ensure it runs only on CPU
 )
 
 # Function to generate comments
 def generate_code_comments(code_snippet):
     prompt = f"### Code:\n{code_snippet}\n### Add meaningful comments to this code:\n"
-    inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=512).to("cuda" if torch.cuda.is_available() else "cpu")
+    inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=512)
     outputs = model.generate(**inputs, max_length=512)
     commented_code = tokenizer.decode(outputs[0], skip_special_tokens=True)
     return commented_code