added RAG

.gitignore

@@ -1 +1,3 @@
-_pycache_/
+_pycache_
+.env
+nohup.out

ai_generate.py

@@ -3,13 +3,57 @@ from openai import OpenAI
 import os
 from transformers import pipeline
 from groq import Groq
+from langchain_community.document_loaders import PyMuPDFLoader
+from langchain_community.document_loaders import TextLoader
+from langchain_community.embeddings.sentence_transformer import (
+    SentenceTransformerEmbeddings,
+)
+from langchain_community.vectorstores import Chroma
+from langchain_text_splitters import CharacterTextSplitter
+from langchain import hub
+from langchain_core.output_parsers import StrOutputParser
+from langchain_core.runnables import RunnablePassthrough
+from langchain.chains import RetrievalQA
+from langchain_groq import ChatGroq
+from dotenv import load_dotenv
+load_dotenv()
 
 groq_client = Groq(
-    api_key=os.environ.get("groq_key"),
+    api_key=os.environ.get("GROQ_API_KEY"),
 )
 
+def create_db_with_langchain(path):
+    loader = PyMuPDFLoader(path)
+    data = loader.load()
+    # split it into chunks
+    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
+    docs = text_splitter.split_documents(data)
+
+    # create the open-source embedding function
+    embedding_function = SentenceTransformerEmbeddings(model_name="all-MiniLM-L6-v2")
+
+    # load it into Chroma
+    db = Chroma.from_documents(docs, embedding_function)
+    return db
+
+
+def generate_groq_rag(text, model, path):
+    llm = ChatGroq(
+        temperature=0,
+        model_name=model,
+    )
+    db = create_db_with_langchain(path)
+    retriever = db.as_retriever(search_type="mmr", search_kwargs={"k": 4, "fetch_k": 20})
+    prompt = hub.pull("rlm/rag-prompt")
+
+    def format_docs(docs):
+        return "\n\n".join(doc.page_content for doc in docs)
+
+    rag_chain = {"context": retriever | format_docs, "question": RunnablePassthrough()} | prompt | llm
+    return rag_chain.invoke(text).content
 
-def generate_groq(text, model):
+
+def generate_groq_base(text, model):
     completion = groq_client.chat.completions.create(
         model=model,
         messages=[
@@ -31,6 +75,12 @@ def generate_groq(text, model):
             response += chunk.choices[0].delta.content or ""
     return response
 
+def generate_groq(text, model, path):
+    if path:
+        return generate_groq_rag(text, model, path)
+    else:
+        return generate_groq_base(text, model)
+
 
 def generate_openai(text, model, openai_client):
     message = [{"role": "user", "content": text}]
@@ -40,15 +90,15 @@ def generate_openai(text, model, openai_client):
     return response.choices[0].message.content
 
 
-def generate(text, model, api):
+def generate(text, model, path, api):
     if model == "Llama 3":
-        return generate_groq(text, "llama3-70b-8192")
+        return generate_groq(text, "llama3-70b-8192", path)
     elif model == "Groq":
-        return generate_groq(text, "llama3-groq-70b-8192-tool-use-preview")
+        return generate_groq(text, "llama3-groq-70b-8192-tool-use-preview", path)
     elif model == "Mistral":
-        return generate_groq(text, "mixtral-8x7b-32768")
+        return generate_groq(text, "mixtral-8x7b-32768", path)
     elif model == "Gemma":
-        return generate_groq(text, "gemma2-9b-it")
+        return generate_groq(text, "gemma2-9b-it", path)
     elif model == "OpenAI GPT 3.5":
         try:
             openai_client = OpenAI(api_key=api)

app.py

@@ -5,7 +5,7 @@ import re
 from humanize import paraphrase_text
 from ai_generate import generate
 import requests
-import language_tool_python
+import language_tool_python 
 import torch
 from gradio_client import Client
 from transformers import GPT2LMHeadModel, GPT2TokenizerFast
@@ -287,6 +287,7 @@ def generate_article(
     ai_model: str,
     content_string: str,
     api_key: str = None,
+    pdf_file_input=None,
     generated_article: str = None,
     user_comments: str = None,
 ) -> str:
@@ -332,7 +333,7 @@ def generate_article(
         )
         article = response.choices[0].message.content.strip()
     else:
-        article = generate(prompt, ai_model, api_key)
+        article = generate(prompt, ai_model, pdf_file_input, api_key)
 
     return clean_text(article)
 
@@ -414,6 +415,7 @@ def generate_and_format(
     month_to,
     day_to,
     domains_to_include,
+    pdf_file_input,
     generated_article: str = None,
     user_comments: str = None,
 ):
@@ -444,6 +446,7 @@ def generate_and_format(
         ai_model,
         content_string,
         api_key,
+        pdf_file_input,
         generated_article,
         user_comments,
     )
@@ -589,11 +592,9 @@ def create_interface():
                         elem_classes="input-highlight-turquoise",
                     )
                     gr.Markdown("# Search Options", elem_classes="text-center text-3xl mb-6")
-                    with gr.Group():
-                        with gr.Row():
-                            google_search_check = gr.Checkbox(
-                                label="Enable Google Search For Recent Sources", value=True
-                            )
+                    with gr.Row():
+                        google_search_check = gr.Checkbox(label="Enable Google Search For Recent Sources", value=True)
+                    with gr.Group(visible=False) as search_options:
                         with gr.Row():
                             month_from = gr.Dropdown(
                                 choices=months,
@@ -621,6 +622,8 @@ def create_interface():
                                 multiselect=True,
                                 label="Domains To Include",
                             )
+                    gr.Markdown("# Add Optional PDF File with Information", elem_classes="text-center text-3xl mb-6")
+                    pdf_file_input = gr.File(label="Upload PDF")
 
                 with gr.Group():
                     gr.Markdown("## AI Model Configuration", elem_classes="text-xl mb-4")
@@ -681,7 +684,7 @@ def create_interface():
                 humanized_output = gr.Markdown(label="Humanized Article", value="\n\n\n\n", render=True)
                 copy_to_input_btn = gr.Button("Copy to Input for AI Check")
 
-        def become_visible(text):
+        def regenerate_visible(text):
             if text:
                 return gr.update(visible=True)
             else:
@@ -693,9 +696,16 @@ def create_interface():
             else:
                 return gr.update(visible=False)
 
+        def search_visible(toggle):
+            if toggle:
+                return gr.update(visible=True)
+            else:
+                return gr.update(visible=False)
+
+        google_search_check.change(search_visible, inputs=google_search_check, outputs=search_options)
         ai_detector_dropdown.change(highlight_visible, inputs=ai_detector_dropdown, outputs=highlighted_text)
-        output_article.change(become_visible, inputs=output_article, outputs=ai_comments)
-        ai_comments.change(become_visible, inputs=output_article, outputs=regenerate_btn)
+        output_article.change(regenerate_visible, inputs=output_article, outputs=ai_comments)
+        ai_comments.change(regenerate_visible, inputs=output_article, outputs=regenerate_btn)
         ai_check_btn.click(highlight_visible, inputs=ai_detector_dropdown, outputs=highlighted_text)
 
         generate_btn.click(
@@ -724,6 +734,7 @@ def create_interface():
                 month_to,
                 day_to,
                 domains_to_include,
+                pdf_file_input,
             ],
             outputs=[output_article],
         )
@@ -754,6 +765,7 @@ def create_interface():
                 month_to,
                 day_to,
                 domains_to_include,
+                pdf_file_input,
                 output_article,
                 ai_comments,
             ],
@@ -779,6 +791,12 @@ def create_interface():
             outputs=[humanized_output],
         )
 
+
+
+
+
+
+
         copy_to_input_btn.click(
             fn=copy_to_input,
             inputs=[humanized_output],
@@ -790,5 +808,5 @@ def create_interface():
 
 if __name__ == "__main__":
     demo = create_interface()
-    demo.launch(server_name="0.0.0.0", share=True, server_port=7890)
-    # demo.launch(server_name="0.0.0.0")
+    # demo.launch(server_name="0.0.0.0", share=True, server_port=7890)
+    demo.launch(server_name="0.0.0.0")

humanize.py

@@ -18,6 +18,8 @@ else:
     print("CUDA is not available. Using CPU instead.")
     device = torch.device("cpu")
 
+
+
 # Configuration for models and their adapters
 model_config = {
     "Base Model": "polygraf-ai/poly-humanizer-base",

requirements.txt

@@ -12,4 +12,11 @@ Unidecode
 BeautifulSoup4
 google-api-python-client
 newspaper3k
-jusText
+jusText
+langchain-groq
+langchainhub
+sentence-transformers
+langchain-community
+pymupdf
+chromadb
+language-tool-python