Spaces:
Sleeping
Sleeping
Gourisankar Padihary
commited on
Commit
·
2889c96
1
Parent(s):
5661370
Capability to modify the llm through UI
Browse files- app.py +48 -2
- config.py +1 -1
- generator/initialize_llm.py +10 -6
- main.py +3 -3
- retriever/retrieve_documents.py +1 -1
app.py
CHANGED
@@ -4,7 +4,8 @@ import threading
|
|
4 |
import time
|
5 |
from generator.compute_metrics import get_attributes_text
|
6 |
from generator.generate_metrics import generate_metrics, retrieve_and_generate_response
|
7 |
-
from config import AppConfig, ConfigConstants
|
|
|
8 |
|
9 |
def launch_gradio(config : AppConfig):
|
10 |
"""
|
@@ -80,17 +81,50 @@ def launch_gradio(config : AppConfig):
|
|
80 |
logging.error(f"Error computing metrics: {e}")
|
81 |
return f"An error occurred: {e}", ""
|
82 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
83 |
# Define Gradio Blocks layout
|
84 |
with gr.Blocks() as interface:
|
85 |
interface.title = "Real Time RAG Pipeline Q&A"
|
86 |
gr.Markdown("### Real Time RAG Pipeline Q&A") # Heading
|
87 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
# Section to display LLM names
|
89 |
with gr.Row():
|
90 |
model_info = f"Embedding Model: {ConfigConstants.EMBEDDING_MODEL_NAME}\n"
|
91 |
model_info += f"Generation LLM: {config.gen_llm.name if hasattr(config.gen_llm, 'name') else 'Unknown'}\n"
|
92 |
model_info += f"Validation LLM: {config.val_llm.name if hasattr(config.val_llm, 'name') else 'Unknown'}\n"
|
93 |
-
gr.Textbox(value=model_info, label="Model Information", interactive=False) # Read-only textbox
|
94 |
|
95 |
# State to store response and source documents
|
96 |
state = gr.State(value={"query": "","response": "", "source_docs": {}})
|
@@ -122,7 +156,19 @@ def launch_gradio(config : AppConfig):
|
|
122 |
inputs=[state],
|
123 |
outputs=[attr_output, metrics_output]
|
124 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
125 |
|
|
|
|
|
|
|
|
|
|
|
|
|
126 |
# Section to display logs
|
127 |
with gr.Row():
|
128 |
start_log_button = gr.Button("Start Log Update", elem_id="start_btn") # Button to start log updates
|
|
|
4 |
import time
|
5 |
from generator.compute_metrics import get_attributes_text
|
6 |
from generator.generate_metrics import generate_metrics, retrieve_and_generate_response
|
7 |
+
from config import AppConfig, ConfigConstants
|
8 |
+
from generator.initialize_llm import initialize_generation_llm, initialize_validation_llm
|
9 |
|
10 |
def launch_gradio(config : AppConfig):
|
11 |
"""
|
|
|
81 |
logging.error(f"Error computing metrics: {e}")
|
82 |
return f"An error occurred: {e}", ""
|
83 |
|
84 |
+
def reinitialize_gen_llm(gen_llm_name):
|
85 |
+
"""Reinitialize the generation LLM and return updated model info."""
|
86 |
+
if gen_llm_name.strip(): # Only update if input is not empty
|
87 |
+
config.gen_llm = initialize_generation_llm(gen_llm_name)
|
88 |
+
|
89 |
+
# Return updated model information
|
90 |
+
updated_model_info = (
|
91 |
+
f"Embedding Model: {ConfigConstants.EMBEDDING_MODEL_NAME}\n"
|
92 |
+
f"Generation LLM: {config.gen_llm.name if hasattr(config.gen_llm, 'name') else 'Unknown'}\n"
|
93 |
+
f"Validation LLM: {config.val_llm.name if hasattr(config.val_llm, 'name') else 'Unknown'}\n"
|
94 |
+
)
|
95 |
+
return updated_model_info
|
96 |
+
|
97 |
+
def reinitialize_val_llm(val_llm_name):
|
98 |
+
"""Reinitialize the generation LLM and return updated model info."""
|
99 |
+
if val_llm_name.strip(): # Only update if input is not empty
|
100 |
+
config.val_llm = initialize_validation_llm(val_llm_name)
|
101 |
+
|
102 |
+
# Return updated model information
|
103 |
+
updated_model_info = (
|
104 |
+
f"Embedding Model: {ConfigConstants.EMBEDDING_MODEL_NAME}\n"
|
105 |
+
f"Generation LLM: {config.gen_llm.name if hasattr(config.gen_llm, 'name') else 'Unknown'}\n"
|
106 |
+
f"Validation LLM: {config.val_llm.name if hasattr(config.val_llm, 'name') else 'Unknown'}\n"
|
107 |
+
)
|
108 |
+
return updated_model_info
|
109 |
+
|
110 |
# Define Gradio Blocks layout
|
111 |
with gr.Blocks() as interface:
|
112 |
interface.title = "Real Time RAG Pipeline Q&A"
|
113 |
gr.Markdown("### Real Time RAG Pipeline Q&A") # Heading
|
114 |
|
115 |
+
# Textbox for new generation LLM name
|
116 |
+
with gr.Row():
|
117 |
+
new_gen_llm_input = gr.Textbox(label="New Generation LLM Name", placeholder="Enter LLM name to update")
|
118 |
+
update_gen_llm_button = gr.Button("Update Generation LLM")
|
119 |
+
new_val_llm_input = gr.Textbox(label="New Validation LLM Name", placeholder="Enter LLM name to update")
|
120 |
+
update_val_llm_button = gr.Button("Update Validation LLM")
|
121 |
+
|
122 |
# Section to display LLM names
|
123 |
with gr.Row():
|
124 |
model_info = f"Embedding Model: {ConfigConstants.EMBEDDING_MODEL_NAME}\n"
|
125 |
model_info += f"Generation LLM: {config.gen_llm.name if hasattr(config.gen_llm, 'name') else 'Unknown'}\n"
|
126 |
model_info += f"Validation LLM: {config.val_llm.name if hasattr(config.val_llm, 'name') else 'Unknown'}\n"
|
127 |
+
model_info_display = gr.Textbox(value=model_info, label="Model Information", interactive=False) # Read-only textbox
|
128 |
|
129 |
# State to store response and source documents
|
130 |
state = gr.State(value={"query": "","response": "", "source_docs": {}})
|
|
|
156 |
inputs=[state],
|
157 |
outputs=[attr_output, metrics_output]
|
158 |
)
|
159 |
+
|
160 |
+
update_gen_llm_button.click(
|
161 |
+
fn=reinitialize_gen_llm,
|
162 |
+
inputs=[new_gen_llm_input],
|
163 |
+
outputs=[model_info_display] # Update the displayed model info
|
164 |
+
)
|
165 |
|
166 |
+
update_val_llm_button.click(
|
167 |
+
fn=reinitialize_val_llm,
|
168 |
+
inputs=[new_val_llm_input],
|
169 |
+
outputs=[model_info_display] # Update the displayed model info
|
170 |
+
)
|
171 |
+
|
172 |
# Section to display logs
|
173 |
with gr.Row():
|
174 |
start_log_button = gr.Button("Start Log Update", elem_id="start_btn") # Button to start log updates
|
config.py
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
|
2 |
class ConfigConstants:
|
3 |
# Constants related to datasets and models
|
4 |
-
DATA_SET_NAMES = ['covidqa', 'cuad'
|
5 |
EMBEDDING_MODEL_NAME = "sentence-transformers/paraphrase-MiniLM-L3-v2"
|
6 |
RE_RANKER_MODEL_NAME = 'cross-encoder/ms-marco-electra-base'
|
7 |
GENERATION_MODEL_NAME = 'mixtral-8x7b-32768'
|
|
|
1 |
|
2 |
class ConfigConstants:
|
3 |
# Constants related to datasets and models
|
4 |
+
DATA_SET_NAMES = ['covidqa', 'cuad']#, 'delucionqa', 'emanual', 'expertqa', 'finqa', 'hagrid', 'hotpotqa', 'msmarco', 'pubmedqa', 'tatqa', 'techqa']
|
5 |
EMBEDDING_MODEL_NAME = "sentence-transformers/paraphrase-MiniLM-L3-v2"
|
6 |
RE_RANKER_MODEL_NAME = 'cross-encoder/ms-marco-electra-base'
|
7 |
GENERATION_MODEL_NAME = 'mixtral-8x7b-32768'
|
generator/initialize_llm.py
CHANGED
@@ -2,18 +2,22 @@ import logging
|
|
2 |
import os
|
3 |
from langchain_groq import ChatGroq
|
4 |
|
5 |
-
|
6 |
-
|
7 |
-
def initialize_generation_llm():
|
8 |
os.environ["GROQ_API_KEY"] = ""
|
9 |
-
|
|
|
10 |
llm = ChatGroq(model=model_name, temperature=0.7)
|
|
|
11 |
logging.info(f'Generation LLM {model_name} initialized')
|
|
|
12 |
return llm
|
13 |
|
14 |
-
def initialize_validation_llm():
|
15 |
os.environ["GROQ_API_KEY"] = ""
|
16 |
-
|
|
|
17 |
llm = ChatGroq(model=model_name, temperature=0.7)
|
|
|
18 |
logging.info(f'Validation LLM {model_name} initialized')
|
|
|
19 |
return llm
|
|
|
2 |
import os
|
3 |
from langchain_groq import ChatGroq
|
4 |
|
5 |
+
def initialize_generation_llm(input_model_name):
|
|
|
|
|
6 |
os.environ["GROQ_API_KEY"] = ""
|
7 |
+
|
8 |
+
model_name = input_model_name
|
9 |
llm = ChatGroq(model=model_name, temperature=0.7)
|
10 |
+
llm.name = model_name
|
11 |
logging.info(f'Generation LLM {model_name} initialized')
|
12 |
+
|
13 |
return llm
|
14 |
|
15 |
+
def initialize_validation_llm(input_model_name):
|
16 |
os.environ["GROQ_API_KEY"] = ""
|
17 |
+
|
18 |
+
model_name = input_model_name
|
19 |
llm = ChatGroq(model=model_name, temperature=0.7)
|
20 |
+
llm.name = model_name
|
21 |
logging.info(f'Validation LLM {model_name} initialized')
|
22 |
+
|
23 |
return llm
|
main.py
CHANGED
@@ -44,10 +44,10 @@ def main():
|
|
44 |
logging.info("Documents embedded")
|
45 |
|
46 |
# Initialize the Generation LLM
|
47 |
-
gen_llm = initialize_generation_llm()
|
48 |
|
49 |
# Initialize the Validation LLM
|
50 |
-
val_llm = initialize_validation_llm()
|
51 |
|
52 |
#Compute RMSE and AUC-ROC for entire dataset
|
53 |
#Enable below code for calculation
|
@@ -55,7 +55,7 @@ def main():
|
|
55 |
#compute_rmse_auc_roc_metrics(gen_llm, val_llm, datasets[data_set_name], vector_store, 10)
|
56 |
|
57 |
# Launch the Gradio app
|
58 |
-
config = AppConfig(vector_store= vector_store, gen_llm= gen_llm, val_llm= val_llm)
|
59 |
launch_gradio(config)
|
60 |
|
61 |
logging.info("Finished!!!")
|
|
|
44 |
logging.info("Documents embedded")
|
45 |
|
46 |
# Initialize the Generation LLM
|
47 |
+
gen_llm = initialize_generation_llm(ConfigConstants.GENERATION_MODEL_NAME)
|
48 |
|
49 |
# Initialize the Validation LLM
|
50 |
+
val_llm = initialize_validation_llm(ConfigConstants.VALIDATION_MODEL_NAME)
|
51 |
|
52 |
#Compute RMSE and AUC-ROC for entire dataset
|
53 |
#Enable below code for calculation
|
|
|
55 |
#compute_rmse_auc_roc_metrics(gen_llm, val_llm, datasets[data_set_name], vector_store, 10)
|
56 |
|
57 |
# Launch the Gradio app
|
58 |
+
config = AppConfig(vector_store= vector_store, gen_llm = gen_llm, val_llm = val_llm)
|
59 |
launch_gradio(config)
|
60 |
|
61 |
logging.info("Finished!!!")
|
retriever/retrieve_documents.py
CHANGED
@@ -8,7 +8,7 @@ def retrieve_top_k_documents(vector_store, query, top_k=5):
|
|
8 |
documents = vector_store.similarity_search(query, k=top_k)
|
9 |
logging.info(f"Top {top_k} documents reterived for query")
|
10 |
|
11 |
-
documents = rerank_documents(query, documents)
|
12 |
|
13 |
return documents
|
14 |
|
|
|
8 |
documents = vector_store.similarity_search(query, k=top_k)
|
9 |
logging.info(f"Top {top_k} documents reterived for query")
|
10 |
|
11 |
+
#documents = rerank_documents(query, documents)
|
12 |
|
13 |
return documents
|
14 |
|