Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -6,10 +6,7 @@ from PyPDF2 import PdfReader
|
|
6 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
7 |
from langchain.callbacks.manager import CallbackManager
|
8 |
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
|
9 |
-
from
|
10 |
-
|
11 |
-
from langchain.vectorstores import Qdrant
|
12 |
-
from transformers import AutoModelForCausalLM
|
13 |
|
14 |
# Load the embedding model
|
15 |
encoder = SentenceTransformer('jinaai/jina-embedding-b-en-v1')
|
@@ -17,8 +14,6 @@ print("Embedding model loaded...")
|
|
17 |
|
18 |
# Load the LLM
|
19 |
callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
|
20 |
-
|
21 |
-
'''
|
22 |
llm = AutoModelForCausalLM.from_pretrained(
|
23 |
"TheBloke/Llama-2-7B-Chat-GGUF",
|
24 |
model_file="llama-2-7b-chat.Q3_K_S.gguf",
|
@@ -27,32 +22,25 @@ llm = AutoModelForCausalLM.from_pretrained(
|
|
27 |
repetition_penalty=1.5,
|
28 |
max_new_tokens=300,
|
29 |
)
|
30 |
-
'''
|
31 |
-
llm = LlamaCpp(
|
32 |
-
model_path="./llama-2-7b-chat.Q3_K_S.gguf",
|
33 |
-
temperature = 0.2,
|
34 |
-
n_ctx=2048,
|
35 |
-
f16_kv=True, # MUST set to True, otherwise you will run into problem after a couple of calls
|
36 |
-
max_tokens = 500,
|
37 |
-
callback_manager=callback_manager,
|
38 |
-
verbose=True,
|
39 |
-
)
|
40 |
print("LLM loaded...")
|
41 |
|
42 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
|
44 |
def setup_database(files):
|
45 |
all_chunks = []
|
46 |
for file in files:
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
text_splitter = RecursiveCharacterTextSplitter(chunk_size=250, chunk_overlap=50, length_function=len)
|
51 |
-
chunks = text_splitter.split_text(text)
|
52 |
all_chunks.extend(chunks)
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
client.recreate_collection(
|
57 |
collection_name="my_facts",
|
58 |
vectors_config=models.VectorParams(
|
@@ -60,51 +48,64 @@ def setup_database(files):
|
|
60 |
distance=models.Distance.COSINE,
|
61 |
),
|
62 |
)
|
63 |
-
|
64 |
-
print("Collection created...")
|
65 |
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
)
|
74 |
-
)
|
75 |
|
76 |
-
|
|
|
|
|
|
|
77 |
|
78 |
-
def
|
|
|
79 |
hits = client.search(
|
80 |
collection_name="my_facts",
|
81 |
query_vector=encoder.encode(question).tolist(),
|
82 |
limit=3
|
83 |
)
|
84 |
|
85 |
-
context = " ".join(hit.payload["
|
86 |
-
|
87 |
-
|
88 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
89 |
return response
|
90 |
|
91 |
-
def chat(messages):
|
92 |
-
if
|
93 |
-
|
94 |
-
|
95 |
-
|
|
|
|
|
|
|
96 |
|
97 |
-
|
98 |
fn=chat,
|
99 |
-
inputs=
|
100 |
-
|
|
|
|
|
|
|
101 |
title="Q&A with PDFs 👩🏻💻📓✍🏻💡",
|
102 |
description="This app facilitates a conversation with PDFs uploaded💡",
|
103 |
theme="soft",
|
|
|
104 |
live=True,
|
105 |
-
allow_flagging=False,
|
106 |
)
|
107 |
|
108 |
-
|
109 |
-
# Add a way to upload and setup the database before starting the chat
|
110 |
-
screen.launch()
|
|
|
6 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
7 |
from langchain.callbacks.manager import CallbackManager
|
8 |
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
|
9 |
+
from ctransformers import AutoModelForCausalLM
|
|
|
|
|
|
|
10 |
|
11 |
# Load the embedding model
|
12 |
encoder = SentenceTransformer('jinaai/jina-embedding-b-en-v1')
|
|
|
14 |
|
15 |
# Load the LLM
|
16 |
callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
|
|
|
|
|
17 |
llm = AutoModelForCausalLM.from_pretrained(
|
18 |
"TheBloke/Llama-2-7B-Chat-GGUF",
|
19 |
model_file="llama-2-7b-chat.Q3_K_S.gguf",
|
|
|
22 |
repetition_penalty=1.5,
|
23 |
max_new_tokens=300,
|
24 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
print("LLM loaded...")
|
26 |
|
27 |
+
def get_chunks(text):
|
28 |
+
text_splitter = RecursiveCharacterTextSplitter(
|
29 |
+
chunk_size=250,
|
30 |
+
chunk_overlap=50,
|
31 |
+
length_function=len,
|
32 |
+
)
|
33 |
+
return text_splitter.split_text(text)
|
34 |
|
35 |
def setup_database(files):
|
36 |
all_chunks = []
|
37 |
for file in files:
|
38 |
+
reader = PdfReader(file)
|
39 |
+
text = "".join(page.extract_text() for page in reader.pages)
|
40 |
+
chunks = get_chunks(text)
|
|
|
|
|
41 |
all_chunks.extend(chunks)
|
42 |
+
|
43 |
+
client = QdrantClient(path="./db")
|
|
|
44 |
client.recreate_collection(
|
45 |
collection_name="my_facts",
|
46 |
vectors_config=models.VectorParams(
|
|
|
48 |
distance=models.Distance.COSINE,
|
49 |
),
|
50 |
)
|
|
|
|
|
51 |
|
52 |
+
records = [
|
53 |
+
models.Record(
|
54 |
+
id=idx,
|
55 |
+
vector=encoder.encode(chunk).tolist(),
|
56 |
+
payload={f"chunk_{idx}": chunk}
|
57 |
+
) for idx, chunk in enumerate(all_chunks)
|
58 |
+
]
|
|
|
|
|
59 |
|
60 |
+
client.upload_records(
|
61 |
+
collection_name="my_facts",
|
62 |
+
records=records,
|
63 |
+
)
|
64 |
|
65 |
+
def answer_question(question):
|
66 |
+
client = QdrantClient(path="./db")
|
67 |
hits = client.search(
|
68 |
collection_name="my_facts",
|
69 |
query_vector=encoder.encode(question).tolist(),
|
70 |
limit=3
|
71 |
)
|
72 |
|
73 |
+
context = " ".join(hit.payload[f"chunk_{hit.id}"] for hit in hits)
|
74 |
+
|
75 |
+
system_prompt = """You are a helpful co-worker, you will use the provided context to answer user questions.
|
76 |
+
Read the given context before answering questions and think step by step. If you cannot answer a user question based on
|
77 |
+
the provided context, inform the user. Do not use any other information for answering user. Provide a detailed answer to the question."""
|
78 |
+
|
79 |
+
B_INST, E_INST = "[INST]", "[/INST]"
|
80 |
+
B_SYS, E_SYS = "<<SYS>>\n", "\n<</SYS>>\n\n"
|
81 |
+
|
82 |
+
instruction = f"Context: {context}\nUser: {question}"
|
83 |
+
prompt_template = f"{B_INST}{B_SYS}{system_prompt}{E_SYS}{instruction}{E_INST}"
|
84 |
+
|
85 |
+
response = llm(prompt_template)
|
86 |
return response
|
87 |
|
88 |
+
def chat(messages, files):
|
89 |
+
if files:
|
90 |
+
setup_database(files)
|
91 |
+
if messages:
|
92 |
+
question = messages[-1]["text"]
|
93 |
+
answer = answer_question(question)
|
94 |
+
messages.append({"text": answer, "is_user": False})
|
95 |
+
return messages
|
96 |
|
97 |
+
interface = gr.Interface(
|
98 |
fn=chat,
|
99 |
+
inputs=[
|
100 |
+
gr.Chatbot(label="Chat"),
|
101 |
+
gr.File(label="Upload PDFs", file_count="multiple")
|
102 |
+
],
|
103 |
+
outputs=gr.Chatbot(label="Chat"),
|
104 |
title="Q&A with PDFs 👩🏻💻📓✍🏻💡",
|
105 |
description="This app facilitates a conversation with PDFs uploaded💡",
|
106 |
theme="soft",
|
107 |
+
share=True,
|
108 |
live=True,
|
|
|
109 |
)
|
110 |
|
111 |
+
interface.launch()
|
|
|
|