Init app

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+chroma filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,6 +1,6 @@
 ---
 title: Datak
-emoji: 🔥
+emoji: 🪲
 colorFrom: indigo
 colorTo: pink
 sdk: gradio

app.py

@@ -0,0 +1,59 @@
+from langchain.embeddings import OpenAIEmbeddings # for creating embeddings
+from langchain.vectorstores import Chroma # for the vectorization part
+from langchain.chains import ConversationalRetrievalChain
+from langchain.llms import OpenAI # the LLM model we'll use (CHatGPT)
+import gradio as gr
+from gradio import inputs, outputs
+from gradio.mix import Parallel
+
+max_sources = 4
+
+embedding = OpenAIEmbeddings()
+vectordb = Chroma(persist_directory="/chroma", embedding_function=embedding)
+pdf_qa = ConversationalRetrievalChain.from_llm(OpenAI(temperature=0.9, model_name="gpt-3.5-turbo"),
+                                    vectordb.as_retriever(), return_source_documents=True)
+
+
+def chat_pdf(query, chat_history=""):
+    result = pdf_qa({"question": query, "chat_history": chat_history})
+    answer = result["answer"]
+    source_docs = result["source_documents"]
+
+    print("source_docs", len(source_docs))
+
+    cleaned_docs = []
+    for doc in source_docs:
+        cleaned_content = doc.page_content
+        metadata_info = f"Metadata: {doc.metadata}\n"
+        cleaned_docs.append(metadata_info + cleaned_content)
+
+    # Pad the outputs to match the number of output components in the Gradio interface
+    padded_outputs = [answer] + cleaned_docs + [""] * (max_sources - len(cleaned_docs))
+    return padded_outputs
+    return [answer] + cleaned_docs
+
+
+def create_outputs(num_sources):
+    outputs = [gr.outputs.Textbox(label="Answer")]
+
+    for i in range(1, num_sources + 1):
+        outputs.append(gr.outputs.Textbox(label=f"Source Document {i}"))
+
+    return outputs
+
+
+iface = gr.Interface(
+    fn=chat_pdf,
+    inputs=[gr.inputs.Textbox(label="Query")],
+    outputs=create_outputs(max_sources),
+    layout="vertical",
+    examples=[
+        ["Give 2 species of fulgoroidea"],
+        ["What colors are found among fulgoroidea?"],
+        ["Why are fulgoroidea so cute?"]
+        # Add more example queries if desired
+    ],
+    css=".answer, .source_documents {width: 45%; float: left; margin-right: 20px;}"
+)
+
+iface.launch(debug=True)

chroma/chroma-collections.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa9c6a5a3cfc099f13e2d6431a6b77129ba4279a75bd0b677ac883daaf98f17b
+size 557

chroma/chroma-embeddings.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc05c6b9445f1cdb2f8c9258adc6b1c91c2438500c14054c24c343767f4d1761
+size 44184

requirements.txt

@@ -0,0 +1,4 @@
+openai
+langchain
+chromadb
+gradio