Vemel commited on
Commit
5939414
·
unverified ·
1 Parent(s): dce91b8

Typing fixes, code cleanup

Browse files
Files changed (2) hide show
  1. app/app.py +28 -23
  2. app/langchain_agent.py +15 -70
app/app.py CHANGED
@@ -1,4 +1,5 @@
1
  import uuid
 
2
 
3
  import gradio as gr
4
  from dotenv import load_dotenv
@@ -9,38 +10,42 @@ from app.utils import timing
9
 
10
 
11
  @timing
12
- def predict_interface(message: str, history=None, user_id=None) -> str:
13
- response = llm_inference(message, history, user_id)
14
  return response["output"]
15
 
16
 
17
- session_id = gr.Textbox(value=str(uuid.uuid4()), type="text", label="session_id")
18
- example_sentences = [
19
- "Recommend me something in Quentin Tarantino reggae style",
20
- "Give me songs with calm and relaxing vibes",
21
- "I want to listen to something like the movie Inception",
22
- "I want music that sounds like Lebron James eating soup",
23
- ]
24
- examples = [[example, f"user_{i}"] for i, example in enumerate(example_sentences)]
25
-
26
- chat = gr.ChatInterface(
27
- predict_interface,
28
- additional_inputs=[session_id],
29
- chatbot=gr.Chatbot(height=600),
30
- textbox=gr.Textbox(placeholder="Ask me for music recommendations!", container=False, scale=7),
31
- description="This AI makes song recommendations based on your music style.",
32
- examples=examples,
33
- title="Persona Music song recommender",
34
- retry_btn="Retry",
35
- clear_btn="Clear",
36
- undo_btn=None,
37
- )
 
 
 
38
 
39
 
40
  def main():
41
  logger = get_logger()
42
  load_dotenv(override=True, verbose=True, dotenv_path=".env")
43
  logger.info("Running chat")
 
44
  chat.queue().launch(server_name="0.0.0.0", server_port=7860, share=False)
45
 
46
 
 
1
  import uuid
2
+ from typing import Any
3
 
4
  import gradio as gr
5
  from dotenv import load_dotenv
 
10
 
11
 
12
  @timing
13
+ def predict_interface(message: str, history: Any = None, user_id: str = "user") -> str:
14
+ response = llm_inference(message, user_id)
15
  return response["output"]
16
 
17
 
18
+ def create_chat() -> gr.ChatInterface:
19
+ session_id = gr.Textbox(value=str(uuid.uuid4()), type="text", label="session_id")
20
+ example_sentences = [
21
+ "Recommend me something in Quentin Tarantino reggae style",
22
+ "Give me songs with calm and relaxing vibes",
23
+ "I want to listen to something like the movie Inception",
24
+ "I want music that sounds like Lebron James eating soup",
25
+ ]
26
+ examples = [[example, f"user_{i}"] for i, example in enumerate(example_sentences)]
27
+
28
+ return gr.ChatInterface(
29
+ predict_interface,
30
+ additional_inputs=[session_id],
31
+ chatbot=gr.Chatbot(height=600),
32
+ textbox=gr.Textbox(
33
+ placeholder="Ask me for music recommendations!", container=False, scale=7
34
+ ),
35
+ description="This AI makes song recommendations based on your music style.",
36
+ examples=examples,
37
+ title="Persona Music song recommender",
38
+ retry_btn="Retry",
39
+ clear_btn="Clear",
40
+ undo_btn=None,
41
+ )
42
 
43
 
44
  def main():
45
  logger = get_logger()
46
  load_dotenv(override=True, verbose=True, dotenv_path=".env")
47
  logger.info("Running chat")
48
+ chat = create_chat()
49
  chat.queue().launch(server_name="0.0.0.0", server_port=7860, share=False)
50
 
51
 
app/langchain_agent.py CHANGED
@@ -1,33 +1,30 @@
1
  import os
2
- import pprint
3
  from collections import defaultdict
4
  from operator import itemgetter
5
  from typing import Any, Dict, List
6
 
7
- from langchain.agents import AgentExecutor, tool
8
  from langchain.agents.format_scratchpad import format_to_openai_functions
9
  from langchain.agents.output_parsers import OpenAIFunctionsAgentOutputParser
10
  from langchain.callbacks import get_openai_callback
11
  from langchain.chains.conversation.memory import ConversationBufferWindowMemory
12
  from langchain.chat_models import ChatOpenAI
13
  from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
14
- from langchain.schema import StrOutputParser
15
- from langchain.schema.runnable import RunnableConfig, RunnableLambda, RunnablePassthrough
16
- from langchain.tools.render import format_tool_to_openai_function
17
- from langfuse.callback import CallbackHandler
18
 
19
  from app.cyanite import free_text_search
20
  from app.logger import get_logger
21
  from app.utils import timing
22
 
23
- system_message = """You are an agent which recommends songs based on music styles provided by the user.
24
- - A music style could be a combination of instruments, genres or sounds.
25
- - Use get_music_style_description to generate a description of the user's music style.
26
- - The styles might contain pop-culture references (artists, movies, TV-Shows, etc) You should include them when generating descriptions.
27
- - Comment on the description of the style and wish the user to enjoy the recommended songs (he will have received them).
28
- - Do not mention any songs or artists, nor give a list of songs.
29
- Write short responses with a respectful and friendly tone.
30
- """
31
 
32
 
33
  describe_music_style_message = """You receive a music style and your goal is to describe it further with genres, instruments and sounds.
@@ -42,8 +39,7 @@ MEMORY_KEY = "history"
42
 
43
  class Globals:
44
  llm_with_tools: Any = None
45
- handlers: List[CallbackHandler] = []
46
- music_styles_to_tracks: Dict[str, List[Any]] = {}
47
  conversation_costs: Dict[str, float] = defaultdict(float)
48
 
49
 
@@ -56,7 +52,7 @@ prompt = ChatPromptTemplate.from_messages(
56
  ]
57
  )
58
 
59
- conversation_memories = defaultdict(
60
  lambda: ConversationBufferWindowMemory(
61
  memory_key=MEMORY_KEY, return_messages=True, output_key="output", k=4
62
  )
@@ -65,41 +61,6 @@ conversation_memories = defaultdict(
65
  # global dicts to store the tracks and the conversation costs
66
 
67
 
68
- @tool
69
- @timing
70
- def get_music_style_description(music_style: str) -> str:
71
- "A tool which describes a music style and returns a description of it"
72
- description = describe_music_style(music_style)
73
- tracks = free_text_search(description, 20)
74
-
75
- # logger = get_logger()
76
- # logger.warning(f"""
77
- # music_style = {music_style}
78
- # music_style_description = {description}
79
- # tracks = {pprint.pformat(tracks)}""")
80
-
81
- # we store the tracks in a global variable so that we can access them later
82
- Globals.music_styles_to_tracks[description] = tracks
83
- # we return only the description to the user
84
- return description
85
-
86
-
87
- @timing
88
- def describe_music_style(music_style: str) -> str:
89
- "A tool used to describe music styles"
90
- llm_describe = ChatOpenAI(temperature=0.0, model="gpt-3.5-turbo")
91
- prompt_describe = ChatPromptTemplate.from_messages(
92
- [
93
- ("system", describe_music_style_message),
94
- ("human", "{music_style}"),
95
- ]
96
- )
97
- runnable = prompt_describe | llm_describe | StrOutputParser()
98
- return runnable.invoke(
99
- {"music_style": music_style}, RunnableConfig(verbose=True, recursion_limit=1)
100
- )
101
-
102
-
103
  @timing
104
  def get_llm_with_tools():
105
  # We instantiate the Chat Model and bind the tool to it.
@@ -124,7 +85,7 @@ def get_llm_with_tools():
124
 
125
 
126
  @timing
127
- def get_agent_executor_from_user_id(user_id) -> AgentExecutor:
128
  "Returns an agent executor for a given user_id"
129
  memory = conversation_memories[user_id]
130
 
@@ -143,7 +104,6 @@ def get_agent_executor_from_user_id(user_id) -> AgentExecutor:
143
 
144
  return AgentExecutor(
145
  agent=agent,
146
- # tools=[get_music_style_description],
147
  tools=[],
148
  memory=memory,
149
  callbacks=Globals.handlers,
@@ -155,22 +115,7 @@ def get_agent_executor_from_user_id(user_id) -> AgentExecutor:
155
 
156
 
157
  @timing
158
- def get_tracks_from_intermediate_steps(intermediate_steps: List) -> List[Dict[str, str]]:
159
- "Given a list of intermediate steps, returns the tracks from the last get_music_style_description action"
160
- if len(intermediate_steps) == 0:
161
- return []
162
-
163
- for action_message, prompt in intermediate_steps[::-1]:
164
- if action_message.tool == "get_music_style_description":
165
- tracks = Globals.music_styles_to_tracks.get(prompt, [])
166
- return tracks
167
-
168
- # if none of the actions is get_music_style_description, return empty list
169
- return []
170
-
171
-
172
- @timing
173
- def llm_inference(message: str, history: Any, user_id: str) -> Dict[str, Any]:
174
  """
175
  This function is called by the API and returns the conversation response
176
  along with the appropriate tracks and costs of the conversation so far
 
1
  import os
 
2
  from collections import defaultdict
3
  from operator import itemgetter
4
  from typing import Any, Dict, List
5
 
6
+ from langchain.agents import AgentExecutor
7
  from langchain.agents.format_scratchpad import format_to_openai_functions
8
  from langchain.agents.output_parsers import OpenAIFunctionsAgentOutputParser
9
  from langchain.callbacks import get_openai_callback
10
  from langchain.chains.conversation.memory import ConversationBufferWindowMemory
11
  from langchain.chat_models import ChatOpenAI
12
  from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
13
+ from langchain.schema.runnable import RunnableLambda, RunnablePassthrough
14
+ from langfuse.callback import BaseCallbackHandler, CallbackHandler
 
 
15
 
16
  from app.cyanite import free_text_search
17
  from app.logger import get_logger
18
  from app.utils import timing
19
 
20
+ # system_message = """You are an agent which recommends songs based on music styles provided by the user.
21
+ # - A music style could be a combination of instruments, genres or sounds.
22
+ # - Use get_music_style_description to generate a description of the user's music style.
23
+ # - The styles might contain pop-culture references (artists, movies, TV-Shows, etc) You should include them when generating descriptions.
24
+ # - Comment on the description of the style and wish the user to enjoy the recommended songs (he will have received them).
25
+ # - Do not mention any songs or artists, nor give a list of songs.
26
+ # Write short responses with a respectful and friendly tone.
27
+ # """
28
 
29
 
30
  describe_music_style_message = """You receive a music style and your goal is to describe it further with genres, instruments and sounds.
 
39
 
40
  class Globals:
41
  llm_with_tools: Any = None
42
+ handlers: List[BaseCallbackHandler] = []
 
43
  conversation_costs: Dict[str, float] = defaultdict(float)
44
 
45
 
 
52
  ]
53
  )
54
 
55
+ conversation_memories: Dict[str, ConversationBufferWindowMemory] = defaultdict(
56
  lambda: ConversationBufferWindowMemory(
57
  memory_key=MEMORY_KEY, return_messages=True, output_key="output", k=4
58
  )
 
61
  # global dicts to store the tracks and the conversation costs
62
 
63
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
64
  @timing
65
  def get_llm_with_tools():
66
  # We instantiate the Chat Model and bind the tool to it.
 
85
 
86
 
87
  @timing
88
+ def get_agent_executor_from_user_id(user_id: str) -> AgentExecutor:
89
  "Returns an agent executor for a given user_id"
90
  memory = conversation_memories[user_id]
91
 
 
104
 
105
  return AgentExecutor(
106
  agent=agent,
 
107
  tools=[],
108
  memory=memory,
109
  callbacks=Globals.handlers,
 
115
 
116
 
117
  @timing
118
+ def llm_inference(message: str, user_id: str) -> Dict[str, Any]:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
119
  """
120
  This function is called by the API and returns the conversation response
121
  along with the appropriate tracks and costs of the conversation so far