code updated.

app.py

@@ -3,7 +3,7 @@ import pandas as pd
 import faiss
 import torch
 import numpy as np
-
+from accelerate import init_empty_weights, load_checkpoint_and_dispatch
 from sentence_transformers import SentenceTransformer
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
@@ -20,13 +20,16 @@ embedding_model = SentenceTransformer("all-MiniLM-L6-v2")
 # ===============================
 # Load LLM on CPU
 # ===============================
-model_id = "PrunaAI/BioMistral-7B-bnb-8bit-smashed"
-tokenizer = AutoTokenizer.from_pretrained("BioMistral/BioMistral-7B")
+model_id = "BioMistral/BioMistral-7B"
+
+print(f"Loading tokenizer and model: {model_id}")
+tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(
     model_id,
-    trust_remote_code=True,
-    device_map=None,  # CPU only
-)
+    torch_dtype=torch.float16,
+    low_cpu_mem_usage=True,
+)# .to("cpu")
+
 tokenizer.pad_token = tokenizer.eos_token
 
 # ===============================