Spaces:
Running
Running
Update main.py
Browse files
main.py
CHANGED
@@ -17,18 +17,31 @@ llm_client = InferenceClient(
|
|
17 |
token=HF_TOKEN,
|
18 |
)
|
19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
# Configure Llama index settings
|
21 |
Settings.llm = HuggingFaceInferenceAPI(
|
22 |
-
model_name=
|
23 |
-
tokenizer_name=
|
24 |
context_window=3000,
|
25 |
token=HF_TOKEN,
|
26 |
max_new_tokens=512,
|
27 |
generate_kwargs={"temperature": 0.1},
|
28 |
)
|
|
|
29 |
Settings.embed_model = HuggingFaceEmbedding(
|
30 |
-
model_name="
|
31 |
-
)
|
32 |
|
33 |
PERSIST_DIR = "db"
|
34 |
PDF_DIRECTORY = 'data'
|
@@ -57,18 +70,42 @@ def data_ingestion_from_directory():
|
|
57 |
index.storage_context.persist(persist_dir=PERSIST_DIR)
|
58 |
|
59 |
def handle_query(query):
|
60 |
-
chat_text_qa_msgs = [
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
72 |
|
73 |
text_qa_template = ChatPromptTemplate.from_messages(chat_text_qa_msgs)
|
74 |
|
|
|
17 |
token=HF_TOKEN,
|
18 |
)
|
19 |
|
20 |
+
# Configure Llama index settings
|
21 |
+
# Settings.llm = HuggingFaceInferenceAPI(
|
22 |
+
# model_name=repo_id,
|
23 |
+
# tokenizer_name=repo_id,
|
24 |
+
# context_window=3000,
|
25 |
+
# token=HF_TOKEN,
|
26 |
+
# max_new_tokens=512,
|
27 |
+
# generate_kwargs={"temperature": 0.1},
|
28 |
+
# )
|
29 |
+
# Settings.embed_model = HuggingFaceEmbedding(
|
30 |
+
# model_name="BAAI/bge-small-en-v1.5"
|
31 |
+
# )
|
32 |
# Configure Llama index settings
|
33 |
Settings.llm = HuggingFaceInferenceAPI(
|
34 |
+
model_name="xlm-roberta-base",
|
35 |
+
tokenizer_name="xlm-roberta-base",
|
36 |
context_window=3000,
|
37 |
token=HF_TOKEN,
|
38 |
max_new_tokens=512,
|
39 |
generate_kwargs={"temperature": 0.1},
|
40 |
)
|
41 |
+
|
42 |
Settings.embed_model = HuggingFaceEmbedding(
|
43 |
+
model_name="sentence-transformers/paraphrase-xlm-r-100langs-v1"
|
44 |
+
)
|
45 |
|
46 |
PERSIST_DIR = "db"
|
47 |
PDF_DIRECTORY = 'data'
|
|
|
70 |
index.storage_context.persist(persist_dir=PERSIST_DIR)
|
71 |
|
72 |
def handle_query(query):
|
73 |
+
# chat_text_qa_msgs = [
|
74 |
+
# (
|
75 |
+
# "user",
|
76 |
+
# """
|
77 |
+
# You are the Taj Hotel chatbot, known as Taj Hotel Helper. Your goal is to provide accurate and professional answers to user queries based on the information available about the Taj Hotel. Always respond clearly and concisely, ideally within 10-15 words. If you don't know the answer, say so politely.
|
78 |
+
# Context:
|
79 |
+
# {context_str}
|
80 |
+
# User's Question:
|
81 |
+
# {query_str}
|
82 |
+
# """
|
83 |
+
# )
|
84 |
+
# ]
|
85 |
+
chat_text_qa_msgs = [
|
86 |
+
(
|
87 |
+
"user",
|
88 |
+
"""
|
89 |
+
You are the Taj Hotel chatbot, known as Taj Hotel Helper.
|
90 |
+
Your goal is to provide accurate and professional answers to
|
91 |
+
user queries about the Taj Hotel in the language they use:
|
92 |
+
English, Telugu, or Hindi. Always respond clearly and concisely,
|
93 |
+
ideally within 10-15 words. If you don't know the answer, say so politely.
|
94 |
+
|
95 |
+
Context:
|
96 |
+
{context_str}
|
97 |
+
|
98 |
+
User's Question:
|
99 |
+
{query_str}
|
100 |
+
|
101 |
+
Language-Specific Guidance:
|
102 |
+
- For English: Respond in English.
|
103 |
+
- For Telugu: తెలుగు లో సమాధానం ఇవ్వండి.
|
104 |
+
- For Hindi: हिंदी में उत्तर दें.
|
105 |
+
"""
|
106 |
+
)
|
107 |
+
]
|
108 |
+
|
109 |
|
110 |
text_qa_template = ChatPromptTemplate.from_messages(chat_text_qa_msgs)
|
111 |
|