ccm commited on
Commit
dc294fb
·
verified ·
1 Parent(s): 81fddf5

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +6 -9
app.py CHANGED
@@ -1,5 +1,4 @@
1
  import threading # to allow streaming response
2
- import time # to pave the deliver of the message
3
 
4
  import gradio # for the interface
5
  import spaces # for GPU
@@ -76,18 +75,16 @@ chatmodel = transformers.AutoModelForCausalLM.from_pretrained(
76
  )
77
 
78
 
79
- def preprocess(query: str, k: int) -> tuple[str, str]:
80
  """
81
  Searches the dataset for the top k most relevant papers to the query and returns a prompt and references
82
  Args:
83
  query (str): The user's query
84
  k (int): The number of results to return
85
  Returns:
86
- tuple[str, str]: A tuple containing the prompt and references
87
  """
88
- documents = publication_vectorstore.search(
89
- query, k=PUBLICATIONS_TO_RETRIEVE, search_type="similarity"
90
- )
91
 
92
  prompt = (
93
  "You are an AI assistant who delights in helping people learn about research. "
@@ -109,7 +106,7 @@ def preprocess(query: str, k: int) -> tuple[str, str]:
109
 
110
  print(prompt)
111
 
112
- return prompt, ""
113
 
114
 
115
  @spaces.GPU
@@ -124,7 +121,7 @@ def reply(message: str, history: list[str]) -> str:
124
  """
125
 
126
  # Apply preprocessing
127
- message, bypass = preprocess(message, PUBLICATIONS_TO_RETRIEVE)
128
 
129
  # This is some handling that is applied to the history variable to put it in a good format
130
  history_transformer_format = [
@@ -150,6 +147,7 @@ def reply(message: str, history: list[str]) -> str:
150
  partial_message += new_token
151
  yield partial_message
152
 
 
153
  # Example queries
154
  EXAMPLE_QUERIES = [
155
  "What is multi-material 3D printing?",
@@ -162,7 +160,6 @@ EXAMPLE_QUERIES = [
162
  "What are the benefits and limitations of using polymers in 3D printing?",
163
  "Tell me about the environmental impacts of additive manufacturing.",
164
  "What are the primary limitations of current 3D printing technologies?",
165
- "What future trends are expected in the field of additive manufacturing?",
166
  "How are researchers improving the speed of 3D printing processes?",
167
  "What are the best practices for managing post-processing in additive manufacturing?",
168
  ]
 
1
  import threading # to allow streaming response
 
2
 
3
  import gradio # for the interface
4
  import spaces # for GPU
 
75
  )
76
 
77
 
78
+ def preprocess(query: str, k: int) -> str:
79
  """
80
  Searches the dataset for the top k most relevant papers to the query and returns a prompt and references
81
  Args:
82
  query (str): The user's query
83
  k (int): The number of results to return
84
  Returns:
85
+ str: The prompt to be used for the AI
86
  """
87
+ documents = publication_vectorstore.search(query, k=k, search_type="similarity")
 
 
88
 
89
  prompt = (
90
  "You are an AI assistant who delights in helping people learn about research. "
 
106
 
107
  print(prompt)
108
 
109
+ return prompt
110
 
111
 
112
  @spaces.GPU
 
121
  """
122
 
123
  # Apply preprocessing
124
+ message = preprocess(message, PUBLICATIONS_TO_RETRIEVE)
125
 
126
  # This is some handling that is applied to the history variable to put it in a good format
127
  history_transformer_format = [
 
147
  partial_message += new_token
148
  yield partial_message
149
 
150
+
151
  # Example queries
152
  EXAMPLE_QUERIES = [
153
  "What is multi-material 3D printing?",
 
160
  "What are the benefits and limitations of using polymers in 3D printing?",
161
  "Tell me about the environmental impacts of additive manufacturing.",
162
  "What are the primary limitations of current 3D printing technologies?",
 
163
  "How are researchers improving the speed of 3D printing processes?",
164
  "What are the best practices for managing post-processing in additive manufacturing?",
165
  ]