🔁 Switched to LaMini-Flan-T5-783M for better CPU reasoning

app.py

@@ -12,10 +12,10 @@ from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_core.documents import Document
 
-# Load Reasoning Model (CPU-friendly + stronger than base)
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
 
-model_id = "google/flan-t5-large"
+# Load Reasoning Model (Balanced for CPU + Reasoning)
+model_id = "MBZUAI/LaMini-Flan-T5-783M"
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
 
@@ -29,6 +29,7 @@ reasoning_pipeline = pipeline(
 )
 
 
+
 # Embedding Model
 embedding_model = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
 vectorstore = None