Spaces:
Runtime error
Runtime error
carolanderson
commited on
Commit
·
7081223
1
Parent(s):
f2f3156
adjust decoding controls
Browse files
app.py
CHANGED
@@ -1,86 +1,76 @@
|
|
1 |
import logging
|
2 |
-
import os
|
3 |
|
4 |
from langchain.chains import LLMChain
|
5 |
from langchain.chat_models import ChatOpenAI
|
6 |
from langchain.llms import HuggingFaceHub
|
7 |
from langchain.prompts.chat import (
|
8 |
-
PromptTemplate,
|
9 |
-
ChatPromptTemplate,
|
10 |
-
MessagesPlaceholder,
|
11 |
-
SystemMessagePromptTemplate,
|
12 |
-
HumanMessagePromptTemplate,
|
13 |
)
|
14 |
from langchain.memory import ConversationBufferWindowMemory
|
15 |
from langchain.memory.chat_message_histories import StreamlitChatMessageHistory
|
16 |
-
from langchain.schema import AIMessage, HumanMessage
|
17 |
from openai.error import AuthenticationError
|
18 |
import streamlit as st
|
19 |
|
20 |
-
from langchain import verbose
|
21 |
-
verbose = True
|
22 |
-
|
23 |
|
24 |
def setup_memory():
|
25 |
msgs = StreamlitChatMessageHistory(key="basic_chat_app")
|
26 |
-
memory = ConversationBufferWindowMemory(k=3, memory_key="chat_history",
|
27 |
chat_memory=msgs,
|
28 |
return_messages=True)
|
29 |
logging.info("setting up new chat memory")
|
30 |
return memory
|
31 |
-
|
32 |
|
33 |
-
|
|
|
34 |
# TODO: consider whether prompt needs to be checked here
|
35 |
if "mistral" in model:
|
36 |
return False
|
37 |
if "current_chain" in st.session_state:
|
38 |
current_chain = st.session_state.current_chain
|
39 |
if (current_chain.model == model) \
|
40 |
-
|
41 |
-
|
42 |
-
and (current_chain.max_tokens == max_tokens):
|
43 |
return True
|
44 |
return False
|
45 |
|
46 |
|
47 |
class CurrentChain():
|
48 |
-
def __init__(self, model, provider, prompt, memory,
|
49 |
self.model = model
|
50 |
self.provider = provider
|
51 |
-
self.
|
52 |
-
|
53 |
-
|
54 |
logging.info(f"setting up new chain with params {model_name}, {provider}, {temp}")
|
55 |
-
if provider == "OpenAI":
|
56 |
-
llm = ChatOpenAI(model_name=model,
|
|
|
|
|
57 |
elif provider == "HuggingFace":
|
58 |
-
# TODO: expose the controls below as widgets and clean up init
|
59 |
llm = HuggingFaceHub(repo_id=model,
|
60 |
-
model_kwargs=
|
61 |
-
|
62 |
-
"top_p" : 0.95,
|
63 |
-
"repetition_penalty" : 1.0,
|
64 |
-
"do_sample" : True,
|
65 |
-
"seed" : 42})
|
66 |
|
67 |
self.conversation = LLMChain(
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
|
74 |
|
75 |
def format_mistral_prompt(message, history):
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
if __name__ == "__main__":
|
85 |
logging.basicConfig(level=logging.INFO)
|
86 |
|
@@ -88,56 +78,53 @@ if __name__ == "__main__":
|
|
88 |
st.write("On small screens, click the `>` at top left to choose options")
|
89 |
with st.expander("How conversation history works"):
|
90 |
st.write("To keep input lengths down and costs reasonable,"
|
91 |
-
"
|
92 |
-
|
|
|
93 |
st.sidebar.title("Choose options")
|
94 |
|
95 |
#### USER INPUT ######
|
96 |
model_name = st.sidebar.selectbox(
|
97 |
-
label
|
98 |
-
options
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
],
|
104 |
help="Which LLM to use",
|
105 |
)
|
106 |
|
107 |
-
st.sidebar.write("Set the decoding temperature. Higher temperatures give "
|
108 |
-
"more unpredictable outputs.")
|
109 |
-
|
110 |
temp = st.sidebar.slider(
|
111 |
label="Temperature",
|
112 |
min_value=float(0),
|
113 |
-
max_value=
|
114 |
step=0.1,
|
115 |
value=0.4,
|
116 |
-
help="Set the decoding temperature"
|
117 |
-
|
118 |
-
|
119 |
-
max_tokens = st.sidebar.slider(
|
120 |
-
label="Max tokens",
|
121 |
-
min_value=32,
|
122 |
-
max_value=2048,
|
123 |
-
step=1,
|
124 |
-
value=1028,
|
125 |
-
help="Set the maximum number of tokens to generate"
|
126 |
-
) # TODO: edit this, not currently using
|
127 |
##########################
|
128 |
-
|
129 |
-
model = model_name.split("(")[0].rstrip()
|
130 |
provider = model_name.split("(")[-1].split(")")[0]
|
131 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
132 |
if "session_memory" not in st.session_state:
|
133 |
st.session_state.session_memory = setup_memory() # for openai
|
134 |
|
135 |
if "history" not in st.session_state:
|
136 |
-
st.session_state.history = []
|
137 |
|
138 |
if "mistral" in model:
|
139 |
-
prompt = PromptTemplate(input_variables=["input"],
|
140 |
-
|
141 |
else:
|
142 |
prompt = ChatPromptTemplate(
|
143 |
messages=[
|
@@ -147,33 +134,32 @@ if __name__ == "__main__":
|
|
147 |
MessagesPlaceholder(variable_name="chat_history"),
|
148 |
HumanMessagePromptTemplate.from_template("{input}")
|
149 |
],
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
if use_existing_chain(model, provider,
|
154 |
chain = st.session_state.current_chain
|
155 |
else:
|
156 |
-
chain =
|
157 |
-
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
162 |
-
st.session_state.current_chain = chain
|
163 |
|
164 |
conversation = chain.conversation
|
165 |
-
|
166 |
if st.button("Clear history"):
|
167 |
-
conversation.memory.clear()
|
168 |
st.session_state.history = [] # for mistral
|
169 |
logging.info("history cleared")
|
170 |
-
|
171 |
for user_msg, asst_msg in st.session_state.history:
|
172 |
with st.chat_message("user"):
|
173 |
st.write(user_msg)
|
174 |
with st.chat_message("assistant"):
|
175 |
st.write(asst_msg)
|
176 |
-
|
177 |
text = st.chat_input()
|
178 |
if text:
|
179 |
with st.chat_message("user"):
|
@@ -192,14 +178,3 @@ if __name__ == "__main__":
|
|
192 |
st.write(result)
|
193 |
except (AuthenticationError, ValueError):
|
194 |
st.warning("Supply a valid API key", icon="⚠️")
|
195 |
-
|
196 |
-
|
197 |
-
|
198 |
-
|
199 |
-
|
200 |
-
|
201 |
-
|
202 |
-
|
203 |
-
|
204 |
-
|
205 |
-
|
|
|
1 |
import logging
|
|
|
2 |
|
3 |
from langchain.chains import LLMChain
|
4 |
from langchain.chat_models import ChatOpenAI
|
5 |
from langchain.llms import HuggingFaceHub
|
6 |
from langchain.prompts.chat import (
|
7 |
+
PromptTemplate,
|
8 |
+
ChatPromptTemplate,
|
9 |
+
MessagesPlaceholder,
|
10 |
+
SystemMessagePromptTemplate,
|
11 |
+
HumanMessagePromptTemplate,
|
12 |
)
|
13 |
from langchain.memory import ConversationBufferWindowMemory
|
14 |
from langchain.memory.chat_message_histories import StreamlitChatMessageHistory
|
|
|
15 |
from openai.error import AuthenticationError
|
16 |
import streamlit as st
|
17 |
|
|
|
|
|
|
|
18 |
|
19 |
def setup_memory():
|
20 |
msgs = StreamlitChatMessageHistory(key="basic_chat_app")
|
21 |
+
memory = ConversationBufferWindowMemory(k=3, memory_key="chat_history",
|
22 |
chat_memory=msgs,
|
23 |
return_messages=True)
|
24 |
logging.info("setting up new chat memory")
|
25 |
return memory
|
|
|
26 |
|
27 |
+
|
28 |
+
def use_existing_chain(model, provider, model_kwargs):
|
29 |
# TODO: consider whether prompt needs to be checked here
|
30 |
if "mistral" in model:
|
31 |
return False
|
32 |
if "current_chain" in st.session_state:
|
33 |
current_chain = st.session_state.current_chain
|
34 |
if (current_chain.model == model) \
|
35 |
+
and (current_chain.provider == provider) \
|
36 |
+
and (current_chain.model_kwargs == model_kwargs):
|
|
|
37 |
return True
|
38 |
return False
|
39 |
|
40 |
|
41 |
class CurrentChain():
|
42 |
+
def __init__(self, model, provider, prompt, memory, model_kwargs):
|
43 |
self.model = model
|
44 |
self.provider = provider
|
45 |
+
self.model_kwargs = model_kwargs
|
46 |
+
|
|
|
47 |
logging.info(f"setting up new chain with params {model_name}, {provider}, {temp}")
|
48 |
+
if provider == "OpenAI":
|
49 |
+
llm = ChatOpenAI(model_name=model,
|
50 |
+
temperature=model_kwargs['temperature']
|
51 |
+
)
|
52 |
elif provider == "HuggingFace":
|
|
|
53 |
llm = HuggingFaceHub(repo_id=model,
|
54 |
+
model_kwargs=model_kwargs
|
55 |
+
)
|
|
|
|
|
|
|
|
|
56 |
|
57 |
self.conversation = LLMChain(
|
58 |
+
llm=llm,
|
59 |
+
prompt=prompt,
|
60 |
+
verbose=True,
|
61 |
+
memory=memory
|
62 |
+
)
|
63 |
|
64 |
|
65 |
def format_mistral_prompt(message, history):
|
66 |
+
prompt = "<s>"
|
67 |
+
for user_prompt, bot_response in history:
|
68 |
+
prompt += f"[INST] {user_prompt} [/INST]"
|
69 |
+
prompt += f" {bot_response}</s> "
|
70 |
+
prompt += f"[INST] {message} [/INST]"
|
71 |
+
return prompt
|
72 |
+
|
73 |
+
|
74 |
if __name__ == "__main__":
|
75 |
logging.basicConfig(level=logging.INFO)
|
76 |
|
|
|
78 |
st.write("On small screens, click the `>` at top left to choose options")
|
79 |
with st.expander("How conversation history works"):
|
80 |
st.write("To keep input lengths down and costs reasonable,"
|
81 |
+
" only the past three turns of conversation "
|
82 |
+
" are used for OpenAI models. Otherwise the entire chat history is used.")
|
83 |
+
st.write("To clear all memory and start fresh, click 'Clear history'")
|
84 |
st.sidebar.title("Choose options")
|
85 |
|
86 |
#### USER INPUT ######
|
87 |
model_name = st.sidebar.selectbox(
|
88 |
+
label="Choose a model",
|
89 |
+
options=["gpt-3.5-turbo (OpenAI)",
|
90 |
+
# "bigscience/bloom (HuggingFace)", # runs
|
91 |
+
# "google/flan-t5-xxl (HuggingFace)", # runs
|
92 |
+
"mistralai/Mistral-7B-Instruct-v0.1 (HuggingFace)"
|
93 |
+
],
|
|
|
94 |
help="Which LLM to use",
|
95 |
)
|
96 |
|
|
|
|
|
|
|
97 |
temp = st.sidebar.slider(
|
98 |
label="Temperature",
|
99 |
min_value=float(0),
|
100 |
+
max_value=2.0,
|
101 |
step=0.1,
|
102 |
value=0.4,
|
103 |
+
help="Set the decoding temperature. "
|
104 |
+
"Higher temps give more unpredictable outputs."
|
105 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
106 |
##########################
|
107 |
+
|
108 |
+
model = model_name.split("(")[0].rstrip() # remove name of model provider
|
109 |
provider = model_name.split("(")[-1].split(")")[0]
|
110 |
|
111 |
+
model_kwargs = {"temperature": temp,
|
112 |
+
"max_new_tokens": 256,
|
113 |
+
"repetition_penalty": 1.0,
|
114 |
+
"top_p": 0.95,
|
115 |
+
"do_sample": True,
|
116 |
+
"seed": 42}
|
117 |
+
# TODO: maybe expose more of these to the user
|
118 |
+
|
119 |
if "session_memory" not in st.session_state:
|
120 |
st.session_state.session_memory = setup_memory() # for openai
|
121 |
|
122 |
if "history" not in st.session_state:
|
123 |
+
st.session_state.history = [] # for mistral
|
124 |
|
125 |
if "mistral" in model:
|
126 |
+
prompt = PromptTemplate(input_variables=["input"],
|
127 |
+
template="{input}")
|
128 |
else:
|
129 |
prompt = ChatPromptTemplate(
|
130 |
messages=[
|
|
|
134 |
MessagesPlaceholder(variable_name="chat_history"),
|
135 |
HumanMessagePromptTemplate.from_template("{input}")
|
136 |
],
|
137 |
+
verbose=True
|
138 |
+
)
|
139 |
+
|
140 |
+
if use_existing_chain(model, provider, model_kwargs):
|
141 |
chain = st.session_state.current_chain
|
142 |
else:
|
143 |
+
chain = CurrentChain(model,
|
144 |
+
provider,
|
145 |
+
prompt,
|
146 |
+
st.session_state.session_memory,
|
147 |
+
model_kwargs)
|
148 |
+
st.session_state.current_chain = chain
|
|
|
149 |
|
150 |
conversation = chain.conversation
|
151 |
+
|
152 |
if st.button("Clear history"):
|
153 |
+
conversation.memory.clear() # for openai
|
154 |
st.session_state.history = [] # for mistral
|
155 |
logging.info("history cleared")
|
156 |
+
|
157 |
for user_msg, asst_msg in st.session_state.history:
|
158 |
with st.chat_message("user"):
|
159 |
st.write(user_msg)
|
160 |
with st.chat_message("assistant"):
|
161 |
st.write(asst_msg)
|
162 |
+
|
163 |
text = st.chat_input()
|
164 |
if text:
|
165 |
with st.chat_message("user"):
|
|
|
178 |
st.write(result)
|
179 |
except (AuthenticationError, ValueError):
|
180 |
st.warning("Supply a valid API key", icon="⚠️")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|