Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,75 +1,50 @@
|
|
1 |
-
import
|
|
|
|
|
|
|
|
|
2 |
|
3 |
-
|
4 |
-
import spaces # for GPU
|
5 |
-
import transformers # to load an LLM
|
6 |
-
import langchain_community.vectorstores # to load the publication vectorstore
|
7 |
-
import langchain_huggingface # for embeddings
|
8 |
-
|
9 |
-
# The greeting message
|
10 |
GREETING = (
|
11 |
-
"Howdy! "
|
12 |
-
"
|
13 |
-
"
|
14 |
-
"
|
|
|
15 |
)
|
16 |
|
17 |
-
#
|
18 |
EMBEDDING_MODEL_NAME = "all-MiniLM-L12-v2"
|
19 |
-
|
20 |
-
# The LLM model name
|
21 |
LLM_MODEL_NAME = "Qwen/Qwen2.5-7B-Instruct"
|
22 |
-
|
23 |
-
# The number of publications to retrieve
|
24 |
PUBLICATIONS_TO_RETRIEVE = 10
|
25 |
|
26 |
|
27 |
-
def embedding(
|
28 |
-
|
29 |
-
device: str = "cuda",
|
30 |
-
normalize_embeddings: bool = False,
|
31 |
-
) -> langchain_huggingface.HuggingFaceEmbeddings:
|
32 |
-
"""
|
33 |
-
Get the embedding function
|
34 |
-
:param model_name: The model name
|
35 |
-
:type model_name: str
|
36 |
-
:param device: The device to use
|
37 |
-
:type device: str
|
38 |
-
:param normalize_embeddings: Whether to normalize embeddings
|
39 |
-
:type normalize_embeddings: bool
|
40 |
-
|
41 |
-
:return: The embedding function
|
42 |
-
:rtype: langchain_huggingface.HuggingFaceEmbeddings
|
43 |
-
"""
|
44 |
return langchain_huggingface.HuggingFaceEmbeddings(
|
45 |
-
model_name=
|
46 |
model_kwargs={"device": device},
|
47 |
encode_kwargs={"normalize_embeddings": normalize_embeddings},
|
48 |
)
|
49 |
|
50 |
|
51 |
def load_publication_vectorstore() -> langchain_community.vectorstores.FAISS:
|
52 |
-
"""
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
|
|
64 |
publication_vectorstore = load_publication_vectorstore()
|
65 |
-
|
66 |
-
|
67 |
-
tokenizer = transformers.AutoTokenizer.from_pretrained(
|
68 |
-
LLM_MODEL_NAME, trust_remote_code=True
|
69 |
-
)
|
70 |
-
streamer = transformers.TextIteratorStreamer(
|
71 |
-
tokenizer, skip_prompt=True, skip_special_tokens=True
|
72 |
-
)
|
73 |
chatmodel = transformers.AutoModelForCausalLM.from_pretrained(
|
74 |
LLM_MODEL_NAME, device_map="auto", torch_dtype="auto", trust_remote_code=True
|
75 |
)
|
@@ -77,78 +52,60 @@ chatmodel = transformers.AutoModelForCausalLM.from_pretrained(
|
|
77 |
|
78 |
def preprocess(query: str, k: int) -> str:
|
79 |
"""
|
80 |
-
|
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 |
-
|
90 |
-
|
91 |
-
|
92 |
-
"
|
93 |
-
"
|
94 |
-
"
|
95 |
-
"===== RESEARCH_EXCERPTS =====:\n{
|
96 |
-
"===== USER_QUERY =====:\n{
|
97 |
"===== ANSWER =====:\n"
|
98 |
)
|
99 |
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
prompt = prompt.replace("{{EXCERPTS_GO_HERE}}", "\n\n".join(research_excerpts))
|
105 |
-
prompt = prompt.replace("{{QUERY_GOES_HERE}}", query)
|
106 |
-
|
107 |
-
print(prompt)
|
108 |
|
|
|
109 |
return prompt
|
110 |
|
111 |
|
112 |
@spaces.GPU
|
113 |
def reply(message: str, history: list[str]) -> str:
|
114 |
"""
|
115 |
-
|
116 |
-
Args:
|
117 |
-
message (str): The user's message
|
118 |
-
history (list[str]): The conversation history
|
119 |
-
Returns:
|
120 |
-
str: The AI's response
|
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 = [
|
128 |
{"role": role, "content": message_pair[idx]}
|
129 |
for message_pair in history
|
130 |
for idx, role in enumerate(["user", "assistant"])
|
131 |
if message_pair[idx] is not None
|
132 |
] + [{"role": "user", "content": message}]
|
133 |
|
134 |
-
#
|
135 |
text = tokenizer.apply_chat_template(
|
136 |
-
|
137 |
)
|
138 |
model_inputs = tokenizer([text], return_tensors="pt").to("cuda")
|
139 |
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
-
|
144 |
-
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
yield partial_message
|
149 |
|
150 |
|
151 |
-
# Example
|
152 |
EXAMPLE_QUERIES = [
|
153 |
"What is multi-material 3D printing?",
|
154 |
"How is additive manufacturing being applied in aerospace?",
|
@@ -164,7 +121,7 @@ EXAMPLE_QUERIES = [
|
|
164 |
"What are the best practices for managing post-processing in additive manufacturing?",
|
165 |
]
|
166 |
|
167 |
-
#
|
168 |
gradio.ChatInterface(
|
169 |
reply,
|
170 |
examples=EXAMPLE_QUERIES,
|
|
|
1 |
+
import gradio # Interface handling
|
2 |
+
import spaces # For GPU
|
3 |
+
import transformers # LLM Loading
|
4 |
+
import langchain_community.vectorstores # Vectorstore for publications
|
5 |
+
import langchain_huggingface # Embeddings
|
6 |
|
7 |
+
# Greeting message
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
GREETING = (
|
9 |
+
"Howdy! I'm an AI agent that uses "
|
10 |
+
"[retrieval-augmented generation](https://en.wikipedia.org/wiki/Retrieval-augmented_generation) "
|
11 |
+
"to answer questions about additive manufacturing research. "
|
12 |
+
"I'm still improving, so bear with me if I make any mistakes. "
|
13 |
+
"What can I help you with today?"
|
14 |
)
|
15 |
|
16 |
+
# Constants
|
17 |
EMBEDDING_MODEL_NAME = "all-MiniLM-L12-v2"
|
|
|
|
|
18 |
LLM_MODEL_NAME = "Qwen/Qwen2.5-7B-Instruct"
|
|
|
|
|
19 |
PUBLICATIONS_TO_RETRIEVE = 10
|
20 |
|
21 |
|
22 |
+
def embedding(device: str = "cuda", normalize_embeddings: bool = False) -> langchain_huggingface.HuggingFaceEmbeddings:
|
23 |
+
"""Loads embedding model with specified device and normalization."""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
return langchain_huggingface.HuggingFaceEmbeddings(
|
25 |
+
model_name=EMBEDDING_MODEL_NAME,
|
26 |
model_kwargs={"device": device},
|
27 |
encode_kwargs={"normalize_embeddings": normalize_embeddings},
|
28 |
)
|
29 |
|
30 |
|
31 |
def load_publication_vectorstore() -> langchain_community.vectorstores.FAISS:
|
32 |
+
"""Load the publication vectorstore safely."""
|
33 |
+
try:
|
34 |
+
return langchain_community.vectorstores.FAISS.load_local(
|
35 |
+
folder_path="publication_vectorstore",
|
36 |
+
embeddings=embedding(),
|
37 |
+
allow_dangerous_deserialization=True,
|
38 |
+
)
|
39 |
+
except Exception as e:
|
40 |
+
print(f"Error loading vectorstore: {e}")
|
41 |
+
return None
|
42 |
+
|
43 |
+
|
44 |
+
# Load vectorstore and models
|
45 |
publication_vectorstore = load_publication_vectorstore()
|
46 |
+
tokenizer = transformers.AutoTokenizer.from_pretrained(LLM_MODEL_NAME, trust_remote_code=True)
|
47 |
+
streamer = transformers.TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
chatmodel = transformers.AutoModelForCausalLM.from_pretrained(
|
49 |
LLM_MODEL_NAME, device_map="auto", torch_dtype="auto", trust_remote_code=True
|
50 |
)
|
|
|
52 |
|
53 |
def preprocess(query: str, k: int) -> str:
|
54 |
"""
|
55 |
+
Generates a prompt based on the top k documents matching the query.
|
|
|
|
|
|
|
|
|
|
|
56 |
"""
|
57 |
documents = publication_vectorstore.search(query, k=k, search_type="similarity")
|
58 |
+
research_excerpts = [f'"... {doc.page_content}..."' for doc in documents]
|
59 |
+
|
60 |
+
# Prompt template
|
61 |
+
prompt_template = (
|
62 |
+
"You are an AI assistant who enjoys helping users learn about research. "
|
63 |
+
"Answer the following question on additive manufacturing research using the RESEARCH_EXCERPTS. "
|
64 |
+
"Provide a concise ANSWER based on these excerpts. Avoid listing references.\n\n"
|
65 |
+
"===== RESEARCH_EXCERPTS =====:\n{research_excerpts}\n\n"
|
66 |
+
"===== USER_QUERY =====:\n{query}\n\n"
|
67 |
"===== ANSWER =====:\n"
|
68 |
)
|
69 |
|
70 |
+
prompt = prompt_template.format(
|
71 |
+
research_excerpts="\n\n".join(research_excerpts), query=query
|
72 |
+
)
|
|
|
|
|
|
|
|
|
|
|
73 |
|
74 |
+
print(prompt) # Useful for debugging prompt content
|
75 |
return prompt
|
76 |
|
77 |
|
78 |
@spaces.GPU
|
79 |
def reply(message: str, history: list[str]) -> str:
|
80 |
"""
|
81 |
+
Generates a response to the user’s message.
|
|
|
|
|
|
|
|
|
|
|
82 |
"""
|
83 |
+
# Preprocess message
|
|
|
84 |
message = preprocess(message, PUBLICATIONS_TO_RETRIEVE)
|
85 |
+
history_formatted = [
|
|
|
|
|
86 |
{"role": role, "content": message_pair[idx]}
|
87 |
for message_pair in history
|
88 |
for idx, role in enumerate(["user", "assistant"])
|
89 |
if message_pair[idx] is not None
|
90 |
] + [{"role": "user", "content": message}]
|
91 |
|
92 |
+
# Tokenize and prepare model input
|
93 |
text = tokenizer.apply_chat_template(
|
94 |
+
history_formatted, tokenize=False, add_generation_prompt=True
|
95 |
)
|
96 |
model_inputs = tokenizer([text], return_tensors="pt").to("cuda")
|
97 |
|
98 |
+
# Generate response directly
|
99 |
+
output_tokens = chatmodel.generate(
|
100 |
+
**model_inputs, max_new_tokens=512
|
101 |
+
)
|
102 |
+
|
103 |
+
# Decode the output tokens
|
104 |
+
response = tokenizer.decode(output_tokens[0], skip_special_tokens=True)
|
105 |
+
return response
|
|
|
106 |
|
107 |
|
108 |
+
# Example Queries for Interface
|
109 |
EXAMPLE_QUERIES = [
|
110 |
"What is multi-material 3D printing?",
|
111 |
"How is additive manufacturing being applied in aerospace?",
|
|
|
121 |
"What are the best practices for managing post-processing in additive manufacturing?",
|
122 |
]
|
123 |
|
124 |
+
# Run the Gradio Interface
|
125 |
gradio.ChatInterface(
|
126 |
reply,
|
127 |
examples=EXAMPLE_QUERIES,
|