initial commit

app.py

@@ -9,72 +9,67 @@ from langchain.embeddings.huggingface import HuggingFaceEmbeddings
 import time
 
 API_URL = "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta"
-# retrieval = Client("https://ishaan-mital-ncert-helper-vector-db.hf.space/--replicas/149bg26k5/")
-
-embed_model_id = 'sentence-transformers/all-MiniLM-L6-v2'
-# device = f'cuda:{cuda.current_device()}' if cuda.is_available() else 'cpu'
-
-embed_model = HuggingFaceEmbeddings(
-    model_name=embed_model_id,
-    # model_kwargs={'device': device},
-    # encode_kwargs={'device': device, 'batch_size': 32}
-)
-
-
-pinecone.init(
-    api_key=os.environ.get('PINECONE_API_KEY'),
-    environment=os.environ.get('PINECONE_ENVIRONMENT')
-)
-
-index_name = 'llama-rag'
-index = pinecone.Index(index_name)
-text_field = 'text'  # field in metadata that contains text content
-docs = [
-    "this is one document",
-    "and another document"
-]
-
-embeddings = embed_model.embed_documents(docs)
-if index_name not in pinecone.list_indexes():
-    pinecone.create_index(
-        index_name,
-        dimension=len(embeddings[0]),
-        metric='cosine'
-    )
-    # wait for index to finish initialization
-    while not pinecone.describe_index(index_name).status['ready']:
-        time.sleep(1)
-vectorstore = Pinecone(
-    index, embed_model.embed_query, text_field
-)
-
-def call_llm_api(input_text):
-    headers = {"Authorization": f"Bearer {os.environ.get('API_KEY')}"}
-    payload = {"input": input_text}
+headers = {"Authorization": f"Bearer {os.environ.get('API_KEY')}"}
+retrieval = Client("https://ishaan-mital-ncert-helper-vector-db.hf.space/--replicas/149bg26k5/")
+
+# embed_model_id = 'sentence-transformers/all-MiniLM-L6-v2'
+# # device = f'cuda:{cuda.current_device()}' if cuda.is_available() else 'cpu'
+
+# embed_model = HuggingFaceEmbeddings(
+#     model_name=embed_model_id,
+#     # model_kwargs={'device': device},
+#     # encode_kwargs={'device': device, 'batch_size': 32}
+# )
+
+
+# pinecone.init(
+#     api_key=os.environ.get('PINECONE_API_KEY'),
+#     environment=os.environ.get('PINECONE_ENVIRONMENT')
+# )
+
+# index_name = 'llama-rag'
+# index = pinecone.Index(index_name)
+# text_field = 'text'  # field in metadata that contains text content
+# docs = [
+#     "this is one document",
+#     "and another document"
+# ]
+
+# embeddings = embed_model.embed_documents(docs)
+# if index_name not in pinecone.list_indexes():
+#     pinecone.create_index(
+#         index_name,
+#         dimension=len(embeddings[0]),
+#         metric='cosine'
+#     )
+#     # wait for index to finish initialization
+#     while not pinecone.describe_index(index_name).status['ready']:
+#         time.sleep(1)
+# vectorstore = Pinecone(
+#     index, embed_model.embed_query, text_field
+# )
+
+def call_llm_api(input_text,context):
+    payload = {"input": {
+                    "question": input_text,
+                    "context": context
+                }
+            }
     response = requests.post(API_URL, headers=headers, json=payload)
     return response.json()  # Adjust as needed based on your API response format
 	
-
-from langchain.llms import Runnable
-
-class APIRunnable(Runnable):
-    def __init__(self, api_func):
-        self.api_func = api_func
-
-    def run(self, input_text):
-        return self.api_func(input_text)
-
-api_runnable = APIRunnable(api_func=call_llm_api)
-
-rag_pipeline = RetrievalQA.from_chain_type(
-    llm=api_runnable, chain_type='stuff',
-    retriever=vectorstore.as_retriever()
-)
+# rag_pipeline = RetrievalQA.from_chain_type(
+#     llm=call_llm_api, chain_type='stuff',
+#     retriever=vectorstore.as_retriever()
+# )
 
 
 def main(question):
-    global chatbot
-    return rag_pipeline(question)
+    # return rag_pipeline(question)
+    context = retrieval.predict(question, api_name = "/predict")
+    answer=call_llm_api(question,context)
+    chatbot = answer[1]
+    return answer[1][0][1]
 
 demo = gr.Interface(main, inputs = "text", outputs = "text")