Spaces:
Runtime error
Runtime error
arjunanand13
commited on
Commit
•
09e2eff
1
Parent(s):
50365d1
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,158 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
import json
|
3 |
+
from torch import cuda, bfloat16
|
4 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline, BitsAndBytesConfig, StoppingCriteria, StoppingCriteriaList
|
5 |
+
from langchain.llms import HuggingFacePipeline
|
6 |
+
import gradio as gr
|
7 |
+
import os
|
8 |
+
import faiss
|
9 |
+
import numpy as np
|
10 |
+
from langchain.embeddings import HuggingFaceEmbeddings
|
11 |
+
|
12 |
+
class Chatbot:
|
13 |
+
def __init__(self):
|
14 |
+
self.HF_TOKEN = os.environ.get("HF_TOKEN", None)
|
15 |
+
self.model_id = "mistralai/Mistral-7B-Instruct-v0.2"
|
16 |
+
self.device = f'cuda:{cuda.current_device()}' if cuda.is_available() else 'cpu'
|
17 |
+
self.bnb_config = BitsAndBytesConfig(
|
18 |
+
load_in_4bit=True,
|
19 |
+
bnb_4bit_quant_type='nf4',
|
20 |
+
bnb_4bit_use_double_quant=True,
|
21 |
+
bnb_4bit_compute_dtype=bfloat16
|
22 |
+
)
|
23 |
+
self.tokenizer = AutoTokenizer.from_pretrained(self.model_id, token=self.HF_TOKEN)
|
24 |
+
self.model = AutoModelForCausalLM.from_pretrained(self.model_id, device_map="auto", token=self.HF_TOKEN, quantization_config=self.bnb_config)
|
25 |
+
self.stop_list = ['\nHuman:', '\n```\n']
|
26 |
+
self.stop_token_ids = [self.tokenizer(x)['input_ids'] for x in self.stop_list]
|
27 |
+
self.stop_token_ids = [torch.LongTensor(x).to(self.device) for x in self.stop_token_ids]
|
28 |
+
self.stopping_criteria = StoppingCriteriaList([self.StopOnTokens()])
|
29 |
+
|
30 |
+
self.generate_text = pipeline(
|
31 |
+
model=self.model,
|
32 |
+
tokenizer=self.tokenizer,
|
33 |
+
return_full_text=True,
|
34 |
+
task='text-generation',
|
35 |
+
temperature=0.1,
|
36 |
+
max_new_tokens=2048,
|
37 |
+
)
|
38 |
+
self.llm = HuggingFacePipeline(pipeline=self.generate_text)
|
39 |
+
|
40 |
+
# Initialize the embedding model
|
41 |
+
self.embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2", model_kwargs={"device": "cuda"})
|
42 |
+
|
43 |
+
try:
|
44 |
+
cpu_index = faiss.read_index('faiss_index_new_model3.index')
|
45 |
+
gpu_resource = faiss.StandardGpuResources()
|
46 |
+
self.vectorstore = faiss.index_cpu_to_gpu(gpu_resource, 0, cpu_index)
|
47 |
+
print("Loaded embedding successfully")
|
48 |
+
except Exception as e:
|
49 |
+
print("FAISS could not be imported or index could not be loaded.")
|
50 |
+
raise e
|
51 |
+
|
52 |
+
self.chain = ConversationalRetrievalChain.from_llm(self.llm, self.vectorstore.as_retriever(), return_source_documents=True)
|
53 |
+
self.chat_history = []
|
54 |
+
|
55 |
+
class StopOnTokens(StoppingCriteria):
|
56 |
+
def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
|
57 |
+
for stop_ids in self.stop_token_ids:
|
58 |
+
if torch.eq(input_ids[0][-len(stop_ids):], stop_ids).all():
|
59 |
+
return True
|
60 |
+
return False
|
61 |
+
|
62 |
+
def format_prompt(self, query):
|
63 |
+
prompt=f"""
|
64 |
+
You are a knowledgeable assistant with access to a comprehensive database.
|
65 |
+
I need you to answer my question and provide related information in a specific format.
|
66 |
+
I have provided four relatable json files , choose the most suitable chunks for answering the query
|
67 |
+
Here's what I need:
|
68 |
+
Include a final answer without additional comments, sign-offs, or extra phrases. Be direct and to the point.
|
69 |
+
|
70 |
+
Here's my question:
|
71 |
+
Query:{query}
|
72 |
+
Solution==>
|
73 |
+
Example1
|
74 |
+
Query: "How to use IPU1_0 instead of A15_0 to process NDK in TDA2x-EVM",
|
75 |
+
Solution: "To use IPU1_0 instead of A15_0 to process NDK in TDA2x-EVM, you need to modify the configuration file of the NDK application. Specifically, change the processor reference from 'A15_0' to 'IPU1_0'.",
|
76 |
+
|
77 |
+
Example2
|
78 |
+
Query: "Can BQ25896 support I2C interface?",
|
79 |
+
Solution: "Yes, the BQ25896 charger supports the I2C interface for communication.",
|
80 |
+
"""
|
81 |
+
return prompt
|
82 |
+
|
83 |
+
def qa_infer(self, query):
|
84 |
+
content = ""
|
85 |
+
formatted_prompt = self.format_prompt(query)
|
86 |
+
|
87 |
+
# Embed the query
|
88 |
+
query_embedding = self.embeddings.embed_query(formatted_prompt)
|
89 |
+
|
90 |
+
# Perform the search
|
91 |
+
distances, indices = self.vectorstore.search(np.array([query_embedding]), k=5)
|
92 |
+
|
93 |
+
# Retrieve the top documents
|
94 |
+
for idx in indices[0]:
|
95 |
+
doc = self.vectorstore.get_document(idx)
|
96 |
+
content += "-" * 50 + "\n"
|
97 |
+
content += doc.page_content + "\n"
|
98 |
+
|
99 |
+
result = self.chain({"question": formatted_prompt, "chat_history": self.chat_history})
|
100 |
+
print(content)
|
101 |
+
print("#" * 100)
|
102 |
+
print(result['answer'])
|
103 |
+
|
104 |
+
output_file = "output.txt"
|
105 |
+
with open(output_file, "w") as f:
|
106 |
+
f.write("Query:\n")
|
107 |
+
f.write(query + "\n\n")
|
108 |
+
f.write("Answer:\n")
|
109 |
+
f.write(result['answer'] + "\n\n")
|
110 |
+
f.write("Source Documents:\n")
|
111 |
+
f.write(content + "\n")
|
112 |
+
|
113 |
+
download_link = f'<a href="file/{output_file}" download>Download Output File</a>'
|
114 |
+
return result['answer'], content, download_link
|
115 |
+
|
116 |
+
def launch_interface(self):
|
117 |
+
css_code = """
|
118 |
+
.gradio-container {
|
119 |
+
background-color: #daccdb;
|
120 |
+
}
|
121 |
+
/* Button styling for all buttons */
|
122 |
+
button {
|
123 |
+
background-color: #927fc7; /* Default color for all other buttons */
|
124 |
+
color: black;
|
125 |
+
border: 1px solid black;
|
126 |
+
padding: 10px;
|
127 |
+
margin-right: 10px;
|
128 |
+
font-size: 16px; /* Increase font size */
|
129 |
+
font-weight: bold; /* Make text bold */
|
130 |
+
}
|
131 |
+
"""
|
132 |
+
EXAMPLES = ["TDA4 product planning and datasheet release progress? ",
|
133 |
+
"I'm using Code Composer Studio 5.4.0.00091 and enabled FPv4SPD16 floating point support for CortexM4 in TDA2. However, after building the project, the .asm file shows --float_support=vfplib instead of FPv4SPD16. Why is this happening?",
|
134 |
+
"Master core in TDA2XX is a15 and in TDA3XX it is m4,so we have to shift all modules that are being used by a15 in TDA2XX to m4 in TDA3xx."]
|
135 |
+
|
136 |
+
|
137 |
+
|
138 |
+
file_path = "ticketNames.txt"
|
139 |
+
|
140 |
+
# Read the file content
|
141 |
+
with open(file_path, "r") as file:
|
142 |
+
content = file.read()
|
143 |
+
ticket_names = json.loads(content)
|
144 |
+
dropdown = gr.Dropdown(label="Sample queries", choices=ticket_names)
|
145 |
+
|
146 |
+
tab1 = gr.Interface(fn=self.qa_infer, inputs=[gr.Textbox(label="QUERY", placeholder ="Enter your query here")], allow_flagging='never', examples=EXAMPLES, cache_examples=False, outputs=[gr.Textbox(label="SOLUTION"), gr.Textbox(label="RELATED QUERIES"), gr.HTML()], css=css_code)
|
147 |
+
tab2 = gr.Interface(fn=self.qa_infer, inputs=[dropdown], allow_flagging='never', outputs=[gr.Textbox(label="SOLUTION"), gr.Textbox(label="RELATED QUERIES"), gr.HTML()], css=css_code)#, title="FAQs")
|
148 |
+
|
149 |
+
|
150 |
+
# # Add dummy outputs to each interface
|
151 |
+
# tab1.outputs = dummy_outputs
|
152 |
+
# tab2.outputs = dummy_outputs
|
153 |
+
|
154 |
+
gr.TabbedInterface([tab1, tab2],["Textbox Input", "FAQs"],title="TI E2E FORUM",css=css_code).launch(debug=True)
|
155 |
+
|
156 |
+
# Instantiate and launch the chatbot
|
157 |
+
chatbot = Chatbot()
|
158 |
+
chatbot.launch_interface()
|