Upload 2 files

app.py

@@ -0,0 +1,25 @@
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+from utils import build_faiss_index, retrieve
+
+# Load documents
+with open("documents/1mg_rag.txt") as f:
+    docs = [line.strip() for line in f if line.strip()]
+
+# Build FAISS index
+index, _ = build_faiss_index(docs)
+
+# Load quantized Mistral 7B
+model_id = "TheBloke/Mistral-7B-Instruct-v0.2-GPTQ"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", trust_remote_code=True)
+
+generator = pipeline("text-generation", model=model, tokenizer=tokenizer)
+
+def answer_question(query):
+    context = "\n".join(retrieve(query, index, docs))
+    prompt = f"[INST] Use the following context to answer the question.\n\nContext:\n{context}\n\nQuestion: {query} [/INST]"
+    result = generator(prompt, max_new_tokens=256, do_sample=True, temperature=0.7)
+    return result[0]['generated_text']
+
+gr.Interface(fn=answer_question, inputs="text", outputs="text", title="Mistral RAG").launch()

utils.py

@@ -0,0 +1,20 @@
+from sentence_transformers import SentenceTransformer
+import faiss
+import numpy as np
+
+# Load MiniLM embedder
+embedder = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
+
+def embed_texts(texts):
+    return embedder.encode(texts, convert_to_tensor=False)
+
+def build_faiss_index(texts):
+    embeddings = embed_texts(texts)
+    index = faiss.IndexFlatL2(embeddings[0].shape[0])
+    index.add(np.array(embeddings))
+    return index, embeddings
+
+def retrieve(query, index, docs, k=3):
+    query_embedding = embed_texts([query])
+    distances, indices = index.search(np.array(query_embedding), k)
+    return [docs[i] for i in indices[0]]