Spaces:
Runtime error
Runtime error
ArunSamespace
commited on
Commit
•
f34fc63
1
Parent(s):
ac0c316
Delete model.py
Browse files
model.py
DELETED
@@ -1,207 +0,0 @@
|
|
1 |
-
import os
|
2 |
-
|
3 |
-
os.environ["GOOGLE_API_KEY"] = "AIzaSyAGoYnNPu__70AId7EJS7F_61i69Qmn-wM"
|
4 |
-
os.environ["OPENAI_API_TYPE"] = "azure"
|
5 |
-
# os.environ["OPENAI_API_VERSION"] = "2023-07-01-preview"
|
6 |
-
# # os.environ["OPENAI_API_KEY"] = "5b624f6b71884a488560a86b1fffbf42"
|
7 |
-
# os.environ["OPENAI_API_KEY"] = "9e337d6696ce4a22a9a1b901e2ebb5fb"
|
8 |
-
import sys
|
9 |
-
|
10 |
-
sys.path.append('../')
|
11 |
-
|
12 |
-
from langchain.chat_models import AzureChatOpenAI, ChatOpenAI
|
13 |
-
from langchain.prompts.chat import (ChatPromptTemplate,
|
14 |
-
HumanMessagePromptTemplate,
|
15 |
-
SystemMessagePromptTemplate)
|
16 |
-
from langchain_google_genai import ChatGoogleGenerativeAI
|
17 |
-
from modules.chat.helpers.embedder import CustomEmbeddings
|
18 |
-
from modules.chat.helpers.search import SimilaritySearch
|
19 |
-
|
20 |
-
embeddings = CustomEmbeddings(
|
21 |
-
model="text-embedding-ada-002",
|
22 |
-
model_url="https://year-embedding-ada-002-aiservices-2136192926.openai.azure.com//openai/deployments/fresh-embedding-ada-002/embeddings?api-version=2023-10-01-preview",
|
23 |
-
api_key="6eed3006cdd3445cb3f422a7358ce461"
|
24 |
-
)
|
25 |
-
vector_store = SimilaritySearch.load_from_disk(
|
26 |
-
embedding_function=embeddings,
|
27 |
-
data_dir="../indexs/text-embedding-ada-002/"
|
28 |
-
# data_dir="../indexs/basic-fno-text-embedding-ada-002/"
|
29 |
-
)
|
30 |
-
|
31 |
-
class Model:
|
32 |
-
def __init__(self, model_name: str, **kwargs) -> None:
|
33 |
-
self.model_name = model_name
|
34 |
-
self.llm = self.load_llm(model_name=model_name, **kwargs)
|
35 |
-
|
36 |
-
def load_llm(self, model_name: str, **kwargs):
|
37 |
-
if self.model_name == "gemini-pro":
|
38 |
-
self.retriever = vector_store.as_retriever(search_kwargs={"k": 2}, search_type="similarity")
|
39 |
-
return ChatGoogleGenerativeAI(model=model_name, temperature=0, max_tokens=4096)
|
40 |
-
elif self.model_name == "gpt-3.5-turbo":
|
41 |
-
self.retriever = vector_store.as_retriever(search_kwargs={"k": 2}, search_type="similarity")
|
42 |
-
return AzureChatOpenAI(
|
43 |
-
deployment_name="latest-gpt-35-turbo-16k",
|
44 |
-
temperature=0,
|
45 |
-
max_tokens=4096,
|
46 |
-
# azure_endpoint="https://high-gpt4-32k-0613-aiservices336365459.openai.azure.com/",
|
47 |
-
openai_api_key="9e337d6696ce4a22a9a1b901e2ebb5fb",
|
48 |
-
# openai_api_base="https://jan-2024-gpt35-turbo16k-aiservices800630185.openai.azure.com/",
|
49 |
-
openai_api_base = "https://fresh-gpt35-turbo-aiservices-2112150452.openai.azure.com/",
|
50 |
-
openai_api_version="2023-07-01-preview"
|
51 |
-
)
|
52 |
-
elif self.model_name == "gpt4":
|
53 |
-
self.retriever = vector_store.as_retriever(search_kwargs={"k": kwargs.get("k", 2)}, search_type="similarity")
|
54 |
-
return AzureChatOpenAI(
|
55 |
-
deployment_name="gpt-4-32k",
|
56 |
-
temperature=0,
|
57 |
-
max_tokens=4096,
|
58 |
-
# azure_endpoint="https://high-gpt4-32k-0613-aiservices336365459.openai.azure.com/",
|
59 |
-
openai_api_key="e91a341abb2f4646ab7b0acd3b9d461e",
|
60 |
-
openai_api_base="https://jan-2024-gpt4-ai-aiservices-1959882301.openai.azure.com/",
|
61 |
-
openai_api_version="2023-07-01-preview"
|
62 |
-
)
|
63 |
-
|
64 |
-
self.retriever = vector_store.as_retriever(search_kwargs={"k": kwargs.get("k", 1)}, search_type="similarity")
|
65 |
-
return ChatOpenAI(
|
66 |
-
model=model_name,
|
67 |
-
openai_api_key="EMPTY",
|
68 |
-
openai_api_base="http://localhost:8000/v1",
|
69 |
-
max_tokens=1024,
|
70 |
-
temperature=0,
|
71 |
-
model_kwargs={"stop": ["<|im_end|>", "Query:", "Question:"], "top_p": 0.95}
|
72 |
-
)
|
73 |
-
|
74 |
-
|
75 |
-
def run_qa_result(self, query: str):
|
76 |
-
support_docs = self.retriever.get_relevant_documents(query)
|
77 |
-
sources = list({d.metadata['source'] for d in support_docs})
|
78 |
-
context = "\n\n".join([f"{i + 1}. {d.page_content}" for i, d in enumerate(support_docs)])
|
79 |
-
return context, sources
|
80 |
-
|
81 |
-
def return_prompt(self, system_prompt: str, query: str, context: str):
|
82 |
-
|
83 |
-
# human_template = "Context:\n\n{context}\n\nQuery: {query}"
|
84 |
-
# human_template = "E-Book:\n\n{context}\n\nQuestion: {query}"
|
85 |
-
|
86 |
-
human_template = "\n\nContext:\n\n{context}\n\nQuestion: {query}"
|
87 |
-
# human_template = "\n\nBook:\n\n{context}\n\nQuestion: {query}"
|
88 |
-
|
89 |
-
messages = []
|
90 |
-
if self.model_name in [
|
91 |
-
"gemini-pro",
|
92 |
-
"TheBloke/Mistral-7B-Instruct-v0.2-AWQ",
|
93 |
-
]:
|
94 |
-
human_template = system_prompt + "\n\n" + human_template
|
95 |
-
human_message_prompt = HumanMessagePromptTemplate.from_template(human_template)
|
96 |
-
messages.append(human_message_prompt)
|
97 |
-
else:
|
98 |
-
system_message_prompt = SystemMessagePromptTemplate.from_template(system_prompt)
|
99 |
-
human_message_prompt = HumanMessagePromptTemplate.from_template(human_template)
|
100 |
-
messages.extend([system_message_prompt, human_message_prompt])
|
101 |
-
|
102 |
-
chat_prompt = ChatPromptTemplate.from_messages(messages)
|
103 |
-
return chat_prompt.format_prompt(context=context, query=query).to_messages()
|
104 |
-
|
105 |
-
def run(self, system_prompt: str, query: str):
|
106 |
-
context, sources = self.run_qa_result(query=query)
|
107 |
-
chat_prompt = self.return_prompt(system_prompt=system_prompt, query=query, context=context)
|
108 |
-
# text = "".join(resp.content for resp in self.llm.stream(chat_prompt))
|
109 |
-
# text += "\nSources: \n" + "\n".join([f"{i + 1}. {d}" for i, d in enumerate(sources)])
|
110 |
-
# return text, sources
|
111 |
-
for resp in self.llm.stream(chat_prompt):
|
112 |
-
yield resp.content.replace("$", "₹")
|
113 |
-
|
114 |
-
yield sources
|
115 |
-
# text = "".join(resp.content for resp in self.llm.stream(chat_prompt))
|
116 |
-
# text += "\nSources: \n" + "\n".join([f"{i + 1}. {d}" for i, d in enumerate(sources)])
|
117 |
-
# return text, sources
|
118 |
-
|
119 |
-
def get_sources(query):
|
120 |
-
results = vector_store.similarity_search_with_relevance_scores(query, k=2)
|
121 |
-
return [
|
122 |
-
{
|
123 |
-
"score": r[-1],
|
124 |
-
"source": r[0].metadata['source']
|
125 |
-
}
|
126 |
-
for r in results
|
127 |
-
]
|
128 |
-
|
129 |
-
if __name__ == "__main__":
|
130 |
-
# model = Model(model_name="phi2")
|
131 |
-
# model = Model(model_name="gpt-3.5-turbo")
|
132 |
-
# model = Model(model_name="gemini-pro")
|
133 |
-
# model = Model(model_name="TheBloke/zephyr-7B-beta-AWQ")
|
134 |
-
# model = Model(model_name="TheBloke/neural-chat-7B-v3-3-AWQ")
|
135 |
-
model = Model(model_name="TheBloke/Mistral-7B-Instruct-v0.2-AWQ")
|
136 |
-
model = Model(model_name="gpt4")
|
137 |
-
model = Model(model_name="gpt-3.5-turbo")
|
138 |
-
|
139 |
-
# query = "what is reliance?"
|
140 |
-
# print("results: ", get_sources(query))
|
141 |
-
|
142 |
-
# query = "explain FNO trading?"
|
143 |
-
# print("results: ", get_sources(query))
|
144 |
-
|
145 |
-
# query="What is FNO trading?"
|
146 |
-
# query = "Describe ITM, ATM and OTM"
|
147 |
-
# query = "give formula to calculate intrinsic value in Put and provide an example"
|
148 |
-
# query = "what is the order of delta, theta, gamma and vega amongst options in a given options chain"
|
149 |
-
# query = "Explain apple stock and nasdaq"
|
150 |
-
|
151 |
-
# query = "generate a table with long and short in F&O instruments"
|
152 |
-
# query = "how can we calculate intrinsic value and time value"
|
153 |
-
# query = "give formula to calculate intrinsic value in Put"
|
154 |
-
|
155 |
-
query = "explain exit from a put trade"
|
156 |
-
#
|
157 |
-
# query = "what will be buying cost if I long tesla CE"
|
158 |
-
|
159 |
-
|
160 |
-
# system_prompt="""Use the following pieces of context to answer the question in detail. Provide example only if it is in provided context and make sure to use them in rupees.""",
|
161 |
-
|
162 |
-
# system_prompt = """Use the following pieces of context to answer the question in detail. Provide example only if it is in context and make sure to use them in ₹.
|
163 |
-
# If you don't know the answer, just say 'Please rephrase the question I am unable to answer'"""
|
164 |
-
|
165 |
-
# system_prompt = """Answer the question using the context. Provide examples only from the context and use only Rupees (₹) in examples. If you don't know the answer, just say 'Please rephrase the question I am unable to answer'"""
|
166 |
-
|
167 |
-
# system_prompt = """Your task is to answer the question using the given context.
|
168 |
-
|
169 |
-
# Follow the below rules while answering the question:
|
170 |
-
# - Only create example using the context
|
171 |
-
# - Use only Rupees '₹' to represent currency.
|
172 |
-
# - If you don't know the answer, just say 'Please rephrase the question I am unable to answer'"""
|
173 |
-
|
174 |
-
# system_prompt = """You are an Indian Stock Market Assistant. Your task is to answer the question using the given context. Only create example from the given context and don't use '$'."""
|
175 |
-
|
176 |
-
# query = "what is reliance?"
|
177 |
-
# query = "what is python?"
|
178 |
-
query = "what is an apple stock and nasdq"
|
179 |
-
query = "Generate a tabular format on playing long and short through options"
|
180 |
-
query = "What is FNO Trading?"
|
181 |
-
|
182 |
-
system_prompt = """Answer the question only from context.
|
183 |
-
Provide examples only from the context.
|
184 |
-
If you don't know the answer, just say 'Please rephrase the question I am unable to answer'"""
|
185 |
-
|
186 |
-
system_prompt = "Answer the question only from the e-book. If it is not sufficient then respond as \"Unknown\""
|
187 |
-
system_prompt = """Use the following pieces of book to answer the question at the end. \nIf you don't know the answer, please think rationally and answer from the book"""
|
188 |
-
# system_prompt = """Answer the question using the context. Provide examples only from the context and use only Rupees (₹) in examples. If you don't know the answer, just say 'Please rephrase the question I am unable to answer'"""
|
189 |
-
|
190 |
-
# system_prompt = """Answer the question from the context. Provide examples only from the context. If you don't know the answer, just say 'Please rephrase the question'"""
|
191 |
-
# system_prompt = """Answer the question from the book. Provide examples only from the book. If you don't know the answer, just say 'Please rephrase the question'"""
|
192 |
-
|
193 |
-
response = model.run(
|
194 |
-
system_prompt=system_prompt,
|
195 |
-
query=query
|
196 |
-
)
|
197 |
-
text = ""
|
198 |
-
for resp in response:
|
199 |
-
if isinstance(resp, list):
|
200 |
-
sources = resp
|
201 |
-
break
|
202 |
-
text += resp
|
203 |
-
|
204 |
-
text = text.split("Question")[0].strip("\n")
|
205 |
-
|
206 |
-
print("text: ", text)
|
207 |
-
open("./text.txt", "w").write(text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|