Test push

app.py

@@ -2,7 +2,7 @@ import gradio as gr
 from huggingface_hub import InferenceClient
 
 """
-For information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
+For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
 """
 client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
@@ -60,5 +60,209 @@ demo = gr.ChatInterface(
 )
 
 
+# Install required packages
+# pip install langchain huggingface_hub pandas faiss-cpu numpy transformers torch accelerate bitsandbytes gradio
+
+import pandas as pd
+import numpy as np
+from langchain.vectorstores import FAISS
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.chains import RetrievalQA
+from langchain.prompts import PromptTemplate
+from langchain.llms import HuggingFacePipeline
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+import torch
+import gradio as gr
+
+def load_embeddings(csv_path):
+    try:
+        df = pd.read_csv(csv_path)
+        
+        if 'Cleaned_Ideas' not in df.columns or 'Embeddings' not in df.columns:
+            raise ValueError("CSV must contain 'Cleaned_Ideas' and 'Embeddings' columns")
+            
+        df['Embeddings'] = df['Embeddings'].apply(
+            lambda x: np.fromstring(
+                x.strip("[]").replace("\n", ""),
+                sep=", ",
+                dtype=np.float32
+            )
+        )
+        
+        expected_dim = 768
+        valid_embeddings = df['Embeddings'].apply(lambda x: len(x) == expected_dim)
+        if not valid_embeddings.all():
+            invalid_count = len(df) - valid_embeddings.sum()
+            raise ValueError(f"{invalid_count} entries have invalid embedding dimensions")
+            
+        return df['Cleaned_Ideas'].tolist(), np.array(df['Embeddings'].tolist())
+        
+    except Exception as e:
+        print(f"Error loading embeddings: {str(e)}")
+        raise
+
+# 1. Load texts and embeddings from CSV
+texts, embeddings = load_embeddings("ideas_with_embeddings.csv")
+
+# 2. Improved embedding model and FAISS configuration
+try:
+    embedding_model = HuggingFaceEmbeddings(
+        model_name="sentence-transformers/all-mpnet-base-v2",
+        model_kwargs={'device': 'cpu'},
+        encode_kwargs={'normalize_embeddings': True}
+    )
+    
+    vector_store = FAISS.from_embeddings(
+        text_embeddings=list(zip(texts, embeddings)),
+        embedding=embedding_model,
+        normalize_L2=True
+    )
+    print(f"FAISS index created with {vector_store.index.ntotal} entries")
+    
+except Exception as e:
+    print(f"Vector store creation failed: {str(e)}")
+    raise
+
+# 3. Using an openly available model (GPT-Neo 1.3B)
+model_name = "EleutherAI/gpt-neo-1.3B"
+try:
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModelForCausalLM.from_pretrained(
+        model_name,
+        device_map="auto",
+        torch_dtype=torch.float16,
+        trust_remote_code=True
+    )
+    print("GPT-Neo 1.3B model loaded successfully.")
+    
+except Exception as e:
+    print(f"Model loading failed: {str(e)}")
+    print("You may need to accept the model license at: https://huggingface.co/EleutherAI/gpt-neo-1.3B")
+    raise
+
+# Text generation pipeline
+pipe = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    max_new_tokens=250,
+    temperature=0.7,
+    top_p=0.9,
+    repetition_penalty=1.1,
+    do_sample=True,
+    return_full_text=False,
+    pad_token_id=tokenizer.eos_token_id
+)
+
+llm = HuggingFacePipeline(pipeline=pipe)
+
+# 4. Enhanced prompt template
+template = """<context>
+{context}
+</context>
+
+<question>
+{question}
+</question>
+
+Analyze the philosophical concept using ONLY the provided context. Follow these steps:
+1. Identify key elements from the context
+2. Explain relationships between concepts
+3. Highlight any mentioned philosophical implications
+
+If context is insufficient, state: "This requires deeper analysis beyond current knowledge." 
+
+Answer:"""
+
+prompt = PromptTemplate(
+    template=template,
+    input_variables=["context", "question"],
+    template_format="f-string"
+)
+
+# 5. Optimized retrieval configuration
+qa_chain = RetrievalQA.from_chain_type(
+    llm=llm,
+    chain_type="stuff",
+    retriever=vector_store.as_retriever(
+        search_type="similarity_score_threshold",
+        search_kwargs={
+            "k": 6,
+            "score_threshold": 0.55,
+            "include_metadata": True
+        }
+    ),
+    return_source_documents=True,
+    chain_type_kwargs={
+        "prompt": prompt,
+        "document_prompt": PromptTemplate(
+            input_variables=["page_content"],
+            template="{page_content}"
+        )
+    },
+    verbose=True
+)
+
+# 6. Query processing function
+def process_query(query):
+    if not query.strip():
+        return "Please enter a valid question", ""
+    
+    try:
+        result = qa_chain({"query": query})
+        response = result['result'].split("Answer:")[-1].strip()
+        
+        sources_info = "\n\n**Sources:**\n"
+        for i, doc in enumerate(result['source_documents'][:3], 1):
+            excerpt = doc.page_content[:200].replace("\n", " ") + "..."
+            score = doc.metadata.get('score', 0)
+            sources_info += f"{i}. {excerpt} (Relevance: {score:.2f})\n\n"
+            
+        return response, sources_info
+    
+    except Exception as e:
+        return f"Error processing request: {str(e)}", ""
+
+# 7. Gradio interface
+def create_gradio_interface():
+    with gr.Blocks(title="Philosophical Concepts RAG v2") as demo:
+        gr.Markdown("# Enhanced Philosophical RAG System")
+        gr.Markdown("Yay Now with improved context understanding and source verification")
+        
+        with gr.Row():
+            with gr.Column(scale=4):
+                query_input = gr.Textbox(
+                    label="Your Question",
+                    placeholder="Ask about philosophical concepts...",
+                    lines=3
+                )
+                submit_btn = gr.Button("Analyze", variant="primary")
+            
+        with gr.Row():
+            with gr.Column(scale=3):
+                response_output = gr.Markdown(label="Analysis")
+            with gr.Column(scale=2):
+                sources_output = gr.Markdown(label="Supporting Context")
+                
+        submit_btn.click(
+            fn=process_query,
+            inputs=[query_input],
+            outputs=[response_output, sources_output]
+        )
+        
+        gr.Markdown("## Example Queries")
+        examples = gr.Examples(
+            examples=[
+                ["Explain the relationship between consciousness and reality using the provided context"],
+                ["How does the concept of 'being' differ across philosophical traditions?"],
+                ["Analyze the arguments for and against epistemological dualism"]
+            ],
+            inputs=query_input
+        )
+        
+    return demo
+
+# 8. Launch the interface
 if __name__ == "__main__":
-    demo.launch()
+    demo = create_gradio_interface()
+    demo.launch(share=True)