attempt citing in prompt

docs/examples/demos/vector_db_qa.ipynb

@@ -25,8 +25,8 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 3,
+   "execution_count": 2,
-   "id": "5c7049db",
+   "id": "0d71cf4f",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -35,8 +35,17 @@
     "text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)\n",
     "texts = text_splitter.split_text(state_of_the_union)\n",
     "\n",
-    "embeddings = OpenAIEmbeddings()\n",
+    "embeddings = OpenAIEmbeddings()"
-    "docsearch = FAISS.from_texts(texts, embeddings)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "id": "5c7049db",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "docsearch = FAISS.from_texts(texts, embeddings, metadatas=[{\"source\": f\"www.{i}.com\"} for i in range(len(texts))])"
    ]
   },
   {
@@ -58,7 +67,7 @@
     {
      "data": {
       "text/plain": [
-       "' The President said that Ketanji Brown Jackson is a consensus builder and has received a broad range of support since she was nominated.'"
+       "' The President said that Judge Ketanji Brown Jackson is \"One of our nation’s top legal minds, who will continue Justice Breyer’s legacy of excellence.\"'"
       ]
      },
      "execution_count": 5,
@@ -96,7 +105,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.6"
+   "version": "3.8.7"
   }
  },
  "nbformat": 4,

langchain/chains/vector_db_qa/base.py

@@ -58,7 +58,10 @@ class VectorDBQA(Chain, BaseModel):
         docs = self.vectorstore.similarity_search(question)
         contexts = []
         for j, doc in enumerate(docs):
-            contexts.append(f"Context {j}:\n{doc.page_content}")
+            context_str = f"Context {j}:\n{doc.page_content}"
+            if doc.metadata is not None:
+                context_str += f"\nSource: {doc.metadata['source']}"
+            contexts.append(context_str)
         # TODO: handle cases where this context is too long.
         answer = llm_chain.predict(question=question, context="\n\n".join(contexts))
         return {self.output_key: answer}

langchain/chains/vector_db_qa/prompt.py

@@ -1,7 +1,12 @@
 # flake8: noqa
 from langchain.prompts import Prompt
 
-prompt_template = """Use the following pieces of context to answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up an answer.
+prompt_template = """Use the following pieces of context to answer the question at the end.
+If you don't know the answer, just say that you don't know, don't try to make up an answer.
+If a source is provided for a particular piece of context, please cite that source when repsonding.
+
+For example:
+"Answer goes here..." - [source goes here]
 
 {context}
 

langchain/docstore/document.py

@@ -1,5 +1,5 @@
 """Interface for interacting with a document."""
-from typing import List
+from typing import List, Optional
 
 from pydantic import BaseModel
 
@@ -10,6 +10,7 @@ class Document(BaseModel):
     page_content: str
     lookup_str: str = ""
     lookup_index = 0
+    metadata: Optional[dict] = None
 
     @property
     def paragraphs(self) -> List[str]:

langchain/vectorstores/faiss.py

@@ -1,5 +1,5 @@
 """Wrapper around FAISS vector database."""
-from typing import Any, Callable, List
+from typing import Any, Callable, List, Optional
 
 import numpy as np
 
@@ -54,7 +54,7 @@ class FAISS(VectorStore):
 
     @classmethod
     def from_texts(
-        cls, texts: List[str], embedding: Embeddings, **kwargs: Any
+        cls, texts: List[str], embedding: Embeddings, metadatas: Optional[List[dict]] = None, **kwargs: Any
     ) -> "FAISS":
         """Construct FAISS wrapper from raw documents.
 
@@ -84,6 +84,8 @@ class FAISS(VectorStore):
         embeddings = embedding.embed_documents(texts)
         index = faiss.IndexFlatL2(len(embeddings[0]))
         index.add(np.array(embeddings, dtype=np.float32))
-        documents = [Document(page_content=text) for text in texts]
+        if metadatas is None:
+            metadatas = [None] * len(texts)
+        documents = [Document(page_content=text, metadata=metadatas[i]) for i, text in enumerate(texts)]
         docstore = InMemoryDocstore({str(i): doc for i, doc in enumerate(documents)})
         return cls(embedding.embed_query, index, docstore)