Update analysis.py

analysis.py

@@ -1,25 +1,20 @@
-from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
-from huggingface_hub import login
-import os
+from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
+import os
 
-# ✅ Get API token from environment variable
 HF_TOKEN = os.getenv("HF_TOKEN")
+MODEL_NAME = "mistralai/Mistral-7B-Instruct-v0.1-GGUF"  # Use a smaller model
 
-# ✅ Authenticate with Hugging Face (without exposing the token in code)
-login(HF_TOKEN)
-
-# ✅ Load Model Efficiently
-MODEL_NAME = "tiiuae/falcon-7b-instruct"
+# Load tokenizer
 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, token=HF_TOKEN)
+
+# Load model (use torch.float16 if on GPU, otherwise use torch.float32 for CPU)
 model = AutoModelForCausalLM.from_pretrained(
     MODEL_NAME,
-    token=HF_TOKEN,
-    device_map="auto",
-    torch_dtype=torch.float16,  
-    load_in_8bit=True  
+    torch_dtype=torch.float32,  # Change to float16 if running on GPU
+    device_map="auto",  # Uses CPU if no GPU is available
+    token=HF_TOKEN
 )
-
 # ✅ Create LLM Pipeline
 llm_pipeline = pipeline("text-generation", model=model, tokenizer=tokenizer, device_map="auto")