Update app.py
Browse files
app.py
CHANGED
|
@@ -40,6 +40,16 @@ YOUTUBE_URL_3 = "https://www.youtube.com/watch?v=vw-KWfKwvTQ"
|
|
| 40 |
|
| 41 |
MODEL_NAME = "gpt-4"
|
| 42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
def invoke(openai_api_key, use_rag, rag_db, prompt):
|
| 44 |
if (openai_api_key == ""):
|
| 45 |
raise gr.Error("OpenAI API Key is required.")
|
|
@@ -75,14 +85,15 @@ def invoke(openai_api_key, use_rag, rag_db, prompt):
|
|
| 75 |
# embedding = OpenAIEmbeddings(disallowed_special = ()),
|
| 76 |
# persist_directory = CHROMA_DIR)
|
| 77 |
# Document retrieval
|
| 78 |
-
vector_db = Chroma(embedding_function = OpenAIEmbeddings(),
|
| 79 |
-
|
| 80 |
-
rag_chain = RetrievalQA.from_chain_type(llm,
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
result = rag_chain({"query": prompt})
|
| 85 |
-
result = result["result"]
|
|
|
|
| 86 |
else:
|
| 87 |
chain = LLMChain(llm = llm, prompt = LLM_CHAIN_PROMPT)
|
| 88 |
result = chain.run({"question": prompt})
|
|
|
|
| 40 |
|
| 41 |
MODEL_NAME = "gpt-4"
|
| 42 |
|
| 43 |
+
def document_retrieval_chroma():
|
| 44 |
+
vector_db = Chroma(embedding_function = OpenAIEmbeddings(),
|
| 45 |
+
persist_directory = CHROMA_DIR)
|
| 46 |
+
rag_chain = RetrievalQA.from_chain_type(llm,
|
| 47 |
+
chain_type_kwargs = {"prompt": RAG_CHAIN_PROMPT},
|
| 48 |
+
retriever = vector_db.as_retriever(search_kwargs = {"k": 3}),
|
| 49 |
+
return_source_documents = True)
|
| 50 |
+
result = rag_chain({"query": prompt})
|
| 51 |
+
return result["result"]
|
| 52 |
+
|
| 53 |
def invoke(openai_api_key, use_rag, rag_db, prompt):
|
| 54 |
if (openai_api_key == ""):
|
| 55 |
raise gr.Error("OpenAI API Key is required.")
|
|
|
|
| 85 |
# embedding = OpenAIEmbeddings(disallowed_special = ()),
|
| 86 |
# persist_directory = CHROMA_DIR)
|
| 87 |
# Document retrieval
|
| 88 |
+
##vector_db = Chroma(embedding_function = OpenAIEmbeddings(),
|
| 89 |
+
## persist_directory = CHROMA_DIR)
|
| 90 |
+
##rag_chain = RetrievalQA.from_chain_type(llm,
|
| 91 |
+
## chain_type_kwargs = {"prompt": RAG_CHAIN_PROMPT},
|
| 92 |
+
## retriever = vector_db.as_retriever(search_kwargs = {"k": 3}),
|
| 93 |
+
## return_source_documents = True)
|
| 94 |
+
##result = rag_chain({"query": prompt})
|
| 95 |
+
##result = result["result"]
|
| 96 |
+
result = document_retrieval_chroma()
|
| 97 |
else:
|
| 98 |
chain = LLMChain(llm = llm, prompt = LLM_CHAIN_PROMPT)
|
| 99 |
result = chain.run({"question": prompt})
|