Upload 2 files

src/doc_qa.py

@@ -0,0 +1,124 @@
+from typing import Optional, List
+
+from langchain.document_loaders import TextLoader  #for textfiles
+from langchain.text_splitter import CharacterTextSplitter #text splitter
+from langchain.embeddings import HuggingFaceEmbeddings #for using HugginFace models
+
+from langchain.document_loaders import UnstructuredPDFLoader  #load pdf
+from langchain.indexes import VectorstoreIndexCreator #vectorize db index with chromadb
+from langchain.chains import RetrievalQA
+from langchain.document_loaders import UnstructuredURLLoader  #load urls into docoument-loader
+from langchain.chains.question_answering import load_qa_chain
+from langchain import HuggingFaceHub
+import os
+from langchain.document_loaders import TextLoader, PyMuPDFLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.llms import HuggingFacePipeline
+from langchain.vectorstores import FAISS
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain import PromptTemplate
+from langchain.chains import LLMChain
+from langchain.base_language import BaseLanguageModel
+from docx import Document
+from langchain.document_loaders import DirectoryLoader
+multi_directory_path=r'tmp/'
+
+from transformers import pipeline
+
+embeddings = HuggingFaceEmbeddings(model_name='sentence-transformers/LaBSE')
+
+
+from langchain_community.document_loaders import TextLoader, PyPDFLoader, Docx2txtLoader
+
+after_rag_template = """Answer the question based only on the following context:
+   {context}
+   Question: {question}
+   """
+#pipe = pipeline("text2text-generation", model="google/flan-t5-large" ,max_new_tokens=100)
+#pipe = pipeline("text2text-generation", model="google/mt5-large" ,max_new_tokens=200)
+from transformers import pipeline, AutoTokenizer, AutoModelForSeq2SeqLM
+#tokenizer = AutoTokenizer.from_pretrained("rinna/bilingual-gpt-neox-4b", use_fast=False)
+#model = AutoModelForSeq2SeqLM.from_pretrained("google/mt5-base")
+# Load model directly
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+#tokenizer = AutoTokenizer.from_pretrained("rinna/bilingual-gpt-neox-4b")
+#model = AutoModelForCausalLM.from_pretrained("rinna/bilingual-gpt-neox-4b")
+#pipe = pipeline("text2text-generation", model="rinna/bilingual-gpt-neox-4b" ,max_new_tokens=200)
+#pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer, max_new_tokens=200)
+
+pipe = pipeline("question-answering", model="deepset/xlm-roberta-base-squad2")
+
+llm = HuggingFacePipeline(pipeline=pipe)
+
+def run_custom_qa(question, retrieved_docs):
+    context = " ".join([doc.page_content for doc in retrieved_docs])
+    output = pipe(question=question, context=context)
+    return output["answer"]
+
+def docs_vector_index():
+    from langchain.document_loaders import DirectoryLoader
+    # Define a directory path
+    directory_path = r"C:\Users\savni\PycharmProjects\DocsSearchEngine\tmp"
+
+    # Create the DirectoryLoader, specifying loaders for each file type
+    loader = DirectoryLoader(
+        directory_path,
+        glob="**/*",  # This pattern loads all files; modify as needed
+
+    )
+    docs = loader.load()
+
+    text_splitter = RecursiveCharacterTextSplitter(
+        chunk_size=1024, chunk_overlap=100, separators=[" ", ",", "\n", "."]
+    )
+    print(docs)
+    docs_chunks = text_splitter.split_documents(docs)
+
+    print(f"docs_chunks length: {len(docs_chunks)}")
+    print('********************docs_chunks',docs_chunks)
+    if len(docs_chunks)>0:
+        db = FAISS.from_documents(docs_chunks, embeddings)
+        return db
+    else:
+        return ''
+
+
+#chain = load_qa_chain(llm, chain_type="stuff")
+
+from langchain.prompts import PromptTemplate
+
+template = """You are an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. Below is some information. 
+{context}
+
+Based on the above information only, answer the below question. 
+
+{question} Be concise."""
+
+prompt = PromptTemplate.from_template(template)
+print(prompt.input_variables)
+
+
+#query_llm = LLMChain(llm=llm, prompt=prompt)
+
+# def doc_qa1(query, db):
+#     similar_doc = db.similarity_search(query, k=2)
+#     doc_c=[]
+#     for c in similar_doc:
+#         doc_c.append(c.page_content)
+#     context=''.join(doc_c)
+#     #response = query_llm.run({"context": context, "question": query})
+#     response = query_llm.run(context=context, question=query)
+#     print('response',response)
+#     return response
+
+def doc_qa(query, db):
+    print("*************************custom qa doc_qa",query)
+    retriever = db.as_retriever()
+    relevant_docs = retriever.get_relevant_documents(query)
+    response=run_custom_qa(query, relevant_docs)
+    print('response', response)
+    return response
+
+
+

src/doc_qa_1.py

@@ -0,0 +1,62 @@
+from langchain.vectorstores import FAISS
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.docstore.document import Document
+from transformers import pipeline
+from langchain.chains.question_answering import load_qa_chain
+import os
+
+# Step 1: Load QA pipeline (don't wrap in HuggingFacePipeline)
+embeddings = HuggingFaceEmbeddings(model_name="intfloat/multilingual-e5-small")
+qa_pipeline = pipeline("question-answering", model="deepset/xlm-roberta-base-squad2")
+multi_directory_path=r'tmp/'
+
+def docs_vector_index():
+    from langchain.document_loaders import DirectoryLoader
+    # Define a directory path
+    directory_path = r"C:\Users\savni\PycharmProjects\DocsSearchEngine\tmp"
+
+    # Create the DirectoryLoader, specifying loaders for each file type
+    loader = DirectoryLoader(
+        directory_path,
+        glob="**/*",  # This pattern loads all files; modify as needed
+
+    )
+    docs = loader.load()
+
+    text_splitter = RecursiveCharacterTextSplitter(
+        chunk_size=1024, chunk_overlap=100, separators=[" ", ",", "\n", "."]
+    )
+    print(docs)
+    docs_chunks = text_splitter.split_documents(docs)
+
+    print(f"docs_chunks length: {len(docs_chunks)}")
+    print('********************docs_chunks',docs_chunks)
+    if len(docs_chunks)>0:
+        db = FAISS.from_documents(docs_chunks, embeddings)
+        return db
+    else:
+        return ''
+
+
+
+def run_custom_qa(question, retrieved_docs):
+    context = " ".join([doc.page_content for doc in retrieved_docs])
+    output = qa_pipeline(question=question, context=context)
+    return output #output["answer"]
+
+# # Step 6: Ask question
+# question = "東京大学はいつ設立されましたか？"
+# relevant_docs = retriever.get_relevant_documents(question)
+# answer = run_custom_qa(question, relevant_docs)
+#
+# print("Answer:", answer)
+
+def doc_qa(query, db):
+    print("*************************custom qa doc_qa",query)
+    retriever = db.as_retriever()
+    relevant_docs = retriever.get_relevant_documents(query)
+    response=run_custom_qa(query, relevant_docs)
+    print('response', response)
+    return response