Merge pull request #30 from manikajay/falcon

Added support for Falcon 40B Instruct BF16 and fixed a bug with FLAN-XXL

kendra_retriever_samples/app.py

@@ -6,6 +6,7 @@
 import kendra_chat_flan_xl as flanxl
 import kendra_chat_flan_xxl as flanxxl
 import kendra_chat_open_ai as openai
+import kendra_chat_falcon_40b as falcon40b
 import kendra_chat_llama_2 as llama2
 
 USER_ICON = "images/user-icon.png"
@@ -16,6 +17,7 @@
     'anthropic': 'Anthropic',
     'flanxl': 'Flan XL',
     'flanxxl': 'Flan XXL',
+    'falcon40b': 'Falcon 40B'
     'llama2' : 'Llama 2'
 }
 
@@ -43,6 +45,9 @@
         elif (sys.argv[1] == 'openai'):
             st.session_state['llm_app'] = openai
             st.session_state['llm_chain'] = openai.build_chain()
+        elif (sys.argv[1] == 'falcon40b'):
+            st.session_state['llm_app'] = falcon40b
+            st.session_state['llm_chain'] = falcon40b.build_chain()
         elif (sys.argv[1] == 'llama2'):
             st.session_state['llm_app'] = llama2
             st.session_state['llm_chain'] = llama2.build_chain()

kendra_retriever_samples/kendra_chat_falcon_40b.py

@@ -0,0 +1,117 @@
+from langchain.retrievers import AmazonKendraRetriever
+from langchain.chains import ConversationalRetrievalChain
+from langchain import SagemakerEndpoint
+from langchain.llms.sagemaker_endpoint import LLMContentHandler
+from langchain.prompts import PromptTemplate
+import sys
+import json
+import os
+
+class bcolors:
+  HEADER = '\033[95m'
+  OKBLUE = '\033[94m'
+  OKCYAN = '\033[96m'
+  OKGREEN = '\033[92m'
+  WARNING = '\033[93m'
+  FAIL = '\033[91m'
+  ENDC = '\033[0m'
+  BOLD = '\033[1m'
+  UNDERLINE = '\033[4m'
+
+MAX_HISTORY_LENGTH = 5
+
+def build_chain():
+  region = os.environ["AWS_REGION"]
+  kendra_index_id = os.environ["KENDRA_INDEX_ID"]
+  endpoint_name = os.environ["FALCON_40B_ENDPOINT"]
+
+  class ContentHandler(LLMContentHandler):
+      content_type = "application/json"
+      accepts = "application/json"
+
+      def transform_input(self, prompt: str, model_kwargs: dict) -> bytes:
+          input_str = json.dumps({"inputs": prompt, "parameters": model_kwargs})
+          print("input_str", input_str)
+          return input_str.encode('utf-8')
+
+      def transform_output(self, output: bytes) -> str:
+          response_json = json.loads(output.read().decode("utf-8"))
+          print(response_json)
+          return response_json[0]["generated_text"]
+
+  content_handler = ContentHandler()
+
+  llm=SagemakerEndpoint(
+          endpoint_name=endpoint_name, 
+          region_name=region, 
+          model_kwargs={
+                        "temperature": 0.8, 
+                        "max_length": 10000, 
+                        "max_new_tokens": 512, 
+                        "do_sample": True, 
+                        "top_p": 0.9,
+                        "repetition_penalty": 1.03,
+                        "stop": ["\nUser:","<|endoftext|>","</s>"]
+                     },
+          content_handler=content_handler
+      )
+
+  retriever = AmazonKendraRetriever(index_id=kendra_index_id)
+
+  prompt_template = """
+  The following is a friendly conversation between a human and an AI. 
+  The AI is talkative and provides lots of specific details from its context.
+  If the AI does not know the answer to a question, it truthfully says it 
+  does not know.
+  {context}
+  Instruction: Based on the above documents, provide a detailed answer for, {question} Answer "don't know" 
+  if not present in the document. 
+  Solution:"""
+  PROMPT = PromptTemplate(
+      template=prompt_template, input_variables=["context", "question"]
+  )
+
+  condense_qa_template = """
+  Given the following conversation and a follow up question, rephrase the follow up question 
+  to be a standalone question.
+
+  Chat History:
+  {chat_history}
+  Follow Up Input: {question}
+  Standalone question:"""
+  standalone_question_prompt = PromptTemplate.from_template(condense_qa_template)
+
+  qa = ConversationalRetrievalChain.from_llm(
+        llm=llm, 
+        retriever=retriever, 
+        condense_question_prompt=standalone_question_prompt, 
+        return_source_documents=True, 
+        combine_docs_chain_kwargs={"prompt":PROMPT})
+  return qa
+
+def run_chain(chain, prompt: str, history=[]):
+  return chain({"question": prompt, "chat_history": history})
+
+if __name__ == "__main__":
+  chat_history = []
+  qa = build_chain()
+  print(bcolors.OKBLUE + "Hello! How can I help you?" + bcolors.ENDC)
+  print(bcolors.OKCYAN + "Ask a question, start a New search: or CTRL-D to exit." + bcolors.ENDC)
+  print(">", end=" ", flush=True)
+  for query in sys.stdin:
+    if (query.strip().lower().startswith("new search:")):
+      query = query.strip().lower().replace("new search:","")
+      chat_history = []
+    elif (len(chat_history) == MAX_HISTORY_LENGTH):
+      chat_history.pop(0)
+    result = run_chain(qa, query, chat_history)
+    chat_history.append((query, result["answer"]))
+    print(bcolors.OKGREEN + result['answer'] + bcolors.ENDC)
+    if 'source_documents' in result:
+      print(bcolors.OKGREEN + 'Sources:')
+      for d in result['source_documents']:
+        print(d.metadata['source'])
+    print(bcolors.ENDC)
+    print(bcolors.OKCYAN + "Ask a question, start a New search: or CTRL-D to exit." + bcolors.ENDC)
+    print(">", end=" ", flush=True)
+  print(bcolors.OKBLUE + "Bye" + bcolors.ENDC)

kendra_retriever_samples/kendra_chat_flan_xxl.py

@@ -31,6 +31,7 @@ class ContentHandler(LLMContentHandler):
 
       def transform_input(self, prompt: str, model_kwargs: dict) -> bytes:
           input_str = json.dumps({"text_inputs": prompt, **model_kwargs})
+          print("input_str", input_str)
           return input_str.encode('utf-8')
 
       def transform_output(self, output: bytes) -> str:

kendra_retriever_samples/kendra_retriever_falcon_40b.py

@@ -0,0 +1,77 @@
+from langchain.retrievers import AmazonKendraRetriever
+from langchain.chains import RetrievalQA
+from langchain import OpenAI
+from langchain.prompts import PromptTemplate
+from langchain import SagemakerEndpoint
+from langchain.llms.sagemaker_endpoint import LLMContentHandler
+import json
+import os
+
+
+def build_chain():
+    region = os.environ["AWS_REGION"]
+    kendra_index_id = os.environ["KENDRA_INDEX_ID"]
+    endpoint_name = os.environ["FALCON_40B_ENDPOINT"]
+
+    class ContentHandler(LLMContentHandler):
+        content_type = "application/json"
+        accepts = "application/json"
+
+        def transform_input(self, prompt: str, model_kwargs: dict) -> bytes:
+            input_str = json.dumps({"inputs": prompt, "parameters": model_kwargs})
+            return input_str.encode('utf-8')
+
+        def transform_output(self, output: bytes) -> str:
+            response_json = json.loads(output.read().decode("utf-8"))
+            print(response_json)
+            return response_json[0]["generated_text"]
+
+    content_handler = ContentHandler()
+
+    llm=SagemakerEndpoint(
+            endpoint_name=endpoint_name, 
+            region_name=region, 
+            model_kwargs={"temperature":1e-10, "min_length": 10000, "max_length": 10000, "max_new_tokens": 100},
+            content_handler=content_handler
+        )
+
+    retriever = AmazonKendraRetriever(index_id=kendra_index_id)
+
+    prompt_template = """
+    The following is a friendly conversation between a human and an AI. 
+    The AI is talkative and provides lots of specific details from its context.
+    If the AI does not know the answer to a question, it truthfully says it 
+    does not know.
+    {context}
+    Instruction: Based on the above documents, provide a detailed answer for, {question} Answer "don't know" 
+    if not present in the document. 
+    Solution:"""
+    PROMPT = PromptTemplate(
+        template=prompt_template, input_variables=["context", "question"]
+    )
+    chain_type_kwargs = {"prompt": PROMPT}
+    qa = RetrievalQA.from_chain_type(
+        llm, 
+        chain_type="stuff", 
+        retriever=retriever, 
+        chain_type_kwargs=chain_type_kwargs,
+        return_source_documents=True
+    )
+    return qa
+
+def run_chain(chain, prompt: str, history=[]):
+    result = chain(prompt)
+    # To make it compatible with chat samples
+    return {
+        "answer": result['result'],
+        "source_documents": result['source_documents']
+    }
+
+if __name__ == "__main__":
+    chain = build_chain()
+    result = run_chain(chain, "What's SageMaker?")
+    print(result['answer'])
+    if 'source_documents' in result:
+        print('Sources:')
+        for d in result['source_documents']:
+          print(d.metadata['source'])

kendra_retriever_samples/kendra_retriever_flan_xxl.py

@@ -23,6 +23,7 @@ def transform_input(self, prompt: str, model_kwargs: dict) -> bytes:
 
         def transform_output(self, output: bytes) -> str:
             response_json = json.loads(output.read().decode("utf-8"))
+            print(response_json)
             return response_json["generated_texts"][0]
 
     content_handler = ContentHandler()