Spaces:
Sleeping
Sleeping
File size: 8,446 Bytes
0545ca0 54c28b8 0545ca0 9e36f19 0545ca0 19b3ae3 0545ca0 bf6bdc0 54c28b8 7eb3ea0 4b8b848 1443530 bf6bdc0 7eb3ea0 ccbc758 0545ca0 7eb3ea0 4b8b848 7eb3ea0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 |
import torch
import json
from torch import cuda, bfloat16
from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline, BitsAndBytesConfig, StoppingCriteria, StoppingCriteriaList
from langchain.llms import HuggingFacePipeline
from langchain.vectorstores import FAISS
from langchain.chains import ConversationalRetrievalChain
import gradio as gr
from langchain.embeddings import HuggingFaceEmbeddings
import os
class Chatbot:
def __init__(self):
self.HF_TOKEN = os.environ.get("HF_TOKEN", None)
self.model_id = "mistralai/Mistral-7B-Instruct-v0.2"
self.device = f'cuda:{cuda.current_device()}' if cuda.is_available() else 'cpu'
self.bnb_config = BitsAndBytesConfig(
load_in_4bit=True,
bnb_4bit_quant_type='nf4',
bnb_4bit_use_double_quant=True,
bnb_4bit_compute_dtype=bfloat16
)
self.tokenizer = AutoTokenizer.from_pretrained(self.model_id, token=self.HF_TOKEN)
self.model = AutoModelForCausalLM.from_pretrained(self.model_id, device_map="auto", token=self.HF_TOKEN, quantization_config=self.bnb_config)
self.stop_list = ['\nHuman:', '\n```\n']
self.stop_token_ids = [self.tokenizer(x)['input_ids'] for x in self.stop_list]
self.stop_token_ids = [torch.LongTensor(x).to(self.device) for x in self.stop_token_ids]
self.stopping_criteria = StoppingCriteriaList([self.StopOnTokens()])
self.generate_text = pipeline(
model=self.model,
tokenizer=self.tokenizer,
return_full_text=True,
task='text-generation',
temperature=0.1,
max_new_tokens=2048,
)
self.llm = HuggingFacePipeline(pipeline=self.generate_text)
try:
self.vectorstore = FAISS.load_local('faiss_index', HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2", model_kwargs={"device": "cuda"}))
# self.vectorstore = FAISS.load_local('faiss_index', HuggingFaceEmbeddings(model_name="flax-sentence-embeddings/all_datasets_v3_MiniLM-L12", model_kwargs={"device": "cuda"}))
print("Loaded embedding successfully")
except ImportError as e:
print("FAISS could not be imported. Make sure FAISS is installed correctly.")
raise e
self.chain = ConversationalRetrievalChain.from_llm(self.llm, self.vectorstore.as_retriever(), return_source_documents=True)
self.chat_history = []
class StopOnTokens(StoppingCriteria):
def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
for stop_ids in self.stop_token_ids:
if torch.eq(input_ids[0][-len(stop_ids):], stop_ids).all():
return True
return False
def format_prompt(self, query):
prompt=f"""
You are a knowledgeable assistant with access to a comprehensive database.
I need you to answer my question and provide related information in a specific format.
I have provided four relatable json files , choose the most suitable chunks for answering the query
Here's what I need:
Include a final answer without additional comments, sign-offs, or extra phrases. Be direct and to the point.
Here's my question:
Query:{query}
Solution==>
Example1
Query: "How to use IPU1_0 instead of A15_0 to process NDK in TDA2x-EVM",
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'.",
Example2
Query: "Can BQ25896 support I2C interface?",
Solution: "Yes, the BQ25896 charger supports the I2C interface for communication.",
"""
return prompt
def qa_infer(self, query):
content = ""
formatted_prompt = self.format_prompt(query)
result = self.chain({"question": formatted_prompt, "chat_history": self.chat_history})
for doc in result['source_documents']:
content += "-" * 50 + "\n"
content += doc.page_content + "\n"
print(content)
print("#" * 100)
print(result['answer'])
output_file = "output.txt"
with open(output_file, "w") as f:
f.write("Query:\n")
f.write(query + "\n\n")
f.write("Answer:\n")
f.write(result['answer'] + "\n\n")
f.write("Source Documents:\n")
f.write(content + "\n")
download_link = f'<a href="file/{output_file}" download>Download Output File</a>'
return result['answer'], content, download_link
def launch_interface(self):
css_code = """
.gradio-container {
background-color: #daccdb;
}
/* Button styling for all buttons */
button {
background-color: #927fc7; /* Default color for all other buttons */
color: black;
border: 1px solid black;
padding: 10px;
margin-right: 10px;
font-size: 16px; /* Increase font size */
font-weight: bold; /* Make text bold */
}
"""
EXAMPLES = ["TDA4 product planning and datasheet release progress? ",
"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?",
"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."]
file_path = "ticketNames.txt"
# Read the file content
with open(file_path, "r") as file:
content = file.read()
ticket_names = json.loads(content)
dropdown = gr.Dropdown(label="Sample queries", choices=ticket_names)
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)
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")
# # Add dummy outputs to each interface
# tab1.outputs = dummy_outputs
# tab2.outputs = dummy_outputs
gr.TabbedInterface([tab1, tab2],["Textbox Input", "FAQs"],title="TI E2E FORUM",css=css_code).launch(debug=True)
# Instantiate and launch the chatbot
chatbot = Chatbot()
chatbot.launch_interface()
"""Single Tab Input Inference"""
# def launch_interface(self):
# css_code = """
# .gradio-container {
# background-color: #daccdb;
# }
# /* Button styling for all buttons */
# button {
# background-color: #927fc7; /* Default color for all other buttons */
# color: black;
# border: 1px solid black;
# padding: 10px;
# margin-right: 10px;
# font-size: 16px; /* Increase font size */
# font-weight: bold; /* Make text bold */
# }
# """
# EXAMPLES = ["TDA4 product planning and datasheet release progress? ",
# "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?",
# "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."]
# demo = 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)
# demo.launch() |