Upload 9 files

.gitattributes

@@ -34,3 +34,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 local_models/llama-2-7b-chat.Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
+PMS_pdfs/DBS[[:space:]]Sustainability[[:space:]]Report[[:space:]]2019[[:space:]]20200306.pdf filter=lfs diff=lfs merge=lfs -text
+PMS_vector_db/PMS_index/index.faiss filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1 @@
+local_models/

PMS_pdfs/DBS Sustainability Report 2019 20200306.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cac737e5f4acb7a922b68a251d21eace4743dd8f5cc6a733e68cbf6f625a65b
+size 7719722

PMS_vector_db/PMS_index/index.faiss

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9799a3e7f37c066b197cb0f6d075c583ab7a307c7266d42625e512000bda56d4
+size 6246445

PMS_vector_db/PMS_index/index.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8d03b619b6ec988fc0ca73c5e97fbf40b93fffb365f1d9c56b88ae4263c11f2
+size 569560

app.py

@@ -0,0 +1,62 @@
+from langchain.llms import CTransformers
+from langchain.document_loaders import PyPDFLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.vectorstores import FAISS
+from langchain.chains import RetrievalQA
+import time
+import gradio as gr
+import os
+
+llm = CTransformers(model= "local_models/llama-2-7b-chat.Q4_K_M.gguf")
+embeddings = HuggingFaceEmbeddings(model_name = 'local_models/embeddings-bge-large/')
+
+
+
+def load_data(dir_path):
+    files = os.listdir(dir_path)
+    data = []
+    for file in files:
+        print(file)
+        loader = PyPDFLoader(dir_path+file)
+        pages = loader.load_and_split()
+        data.extend(pages)
+    return data
+
+def build_vector_db(data):
+    text_splitter = RecursiveCharacterTextSplitter(
+        # Set a really small chunk size, just to show.
+        chunk_size = 300,
+        chunk_overlap  = 30,
+        length_function = len,
+    )
+    text_chunks = text_splitter.split_documents(data)
+    print(len(text_chunks))
+    docsearch = FAISS.from_documents(text_chunks, embeddings)
+    docsearch.save_local('PMS_vector_db/PMS_index')
+    return docsearch
+
+def get_vector_db(db_path):
+    if os.path.exists(db_path):
+        vector_db = FAISS.load_local(db_path, embeddings)
+        print('loading from the existing vectorDB')
+    else:
+        data = load_data("PMS_pdfs/")
+        vector_db = build_vector_db(data)
+    return vector_db
+
+def predict(prompt,history):
+    vector_db = get_vector_db('PMS_vector_db/PMS_index/')
+    qa = RetrievalQA.from_chain_type(llm=llm, chain_type='stuff',
+                                retriever = vector_db.as_retriever(),
+                                return_source_documents = True)
+    response = qa({'query':prompt})
+    response = response['result']
+    for i in range(len(response)):
+      time.sleep(0.05)
+      yield response[:i+1]
+
+
+gr.ChatInterface(predict).queue().launch()
+
+

requirements.txt

@@ -0,0 +1,5 @@
+langchain
+pypdf
+ctransformers
+faiss-cpu
+gradio