Spaces:
Build error
Build error
Commit
·
3b3c852
1
Parent(s):
f790556
added debugging features
Browse files
app.py
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
from copy import deepcopy
|
|
|
|
| 2 |
import streamlit as st
|
| 3 |
import pandas as pd
|
| 4 |
from io import StringIO
|
|
@@ -8,7 +9,13 @@ import weaviate
|
|
| 8 |
from weaviate.embedded import EmbeddedOptions
|
| 9 |
from weaviate import Client
|
| 10 |
from weaviate.util import generate_uuid5
|
|
|
|
| 11 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
# Initialize TAPAS model and tokenizer
|
| 13 |
tokenizer = AutoTokenizer.from_pretrained("google/tapas-large-finetuned-wtq")
|
| 14 |
model = AutoModelForTableQuestionAnswering.from_pretrained("google/tapas-large-finetuned-wtq")
|
|
@@ -18,6 +25,22 @@ client = weaviate.Client(
|
|
| 18 |
embedded_options=EmbeddedOptions()
|
| 19 |
)
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
# Function to check if a class already exists in Weaviate
|
| 22 |
def class_exists(class_name):
|
| 23 |
try:
|
|
@@ -76,6 +99,8 @@ def ingest_data_to_weaviate(dataframe, class_name, class_description):
|
|
| 76 |
}
|
| 77 |
client.data_object.create(obj)
|
| 78 |
|
|
|
|
|
|
|
| 79 |
|
| 80 |
def query_weaviate(question):
|
| 81 |
# This is a basic example; adapt the query based on the question
|
|
@@ -87,10 +112,12 @@ def ask_llm_chunk(chunk, questions):
|
|
| 87 |
try:
|
| 88 |
inputs = tokenizer(table=chunk, queries=questions, padding="max_length", truncation=True, return_tensors="pt")
|
| 89 |
except Exception as e:
|
|
|
|
| 90 |
st.write(f"An error occurred: {e}")
|
| 91 |
return ["Error occurred while tokenizing"] * len(questions)
|
| 92 |
|
| 93 |
if inputs["input_ids"].shape[1] > 512:
|
|
|
|
| 94 |
st.warning("Token limit exceeded for chunk")
|
| 95 |
return ["Token limit exceeded for chunk"] * len(questions)
|
| 96 |
|
|
@@ -106,13 +133,11 @@ def ask_llm_chunk(chunk, questions):
|
|
| 106 |
if len(coordinates) == 1:
|
| 107 |
row, col = coordinates[0]
|
| 108 |
try:
|
| 109 |
-
st.write(f"DataFrame shape: {chunk.shape}") # Debugging line
|
| 110 |
-
st.write(f"DataFrame columns: {chunk.columns}") # Debugging line
|
| 111 |
-
st.write(f"Trying to access row {row}, col {col}") # Debugging line
|
| 112 |
value = chunk.iloc[row, col]
|
| 113 |
-
|
| 114 |
answers.append(value)
|
| 115 |
except Exception as e:
|
|
|
|
| 116 |
st.write(f"An error occurred: {e}")
|
| 117 |
else:
|
| 118 |
cell_values = []
|
|
@@ -122,6 +147,7 @@ def ask_llm_chunk(chunk, questions):
|
|
| 122 |
value = chunk.iloc[row, col]
|
| 123 |
cell_values.append(value)
|
| 124 |
except Exception as e:
|
|
|
|
| 125 |
st.write(f"An error occurred: {e}")
|
| 126 |
answers.append(", ".join(map(str, cell_values)))
|
| 127 |
|
|
@@ -180,6 +206,9 @@ if selected_class != "New Class":
|
|
| 180 |
if csv_file is not None:
|
| 181 |
data = csv_file.read().decode("utf-8")
|
| 182 |
dataframe = pd.read_csv(StringIO(data))
|
|
|
|
|
|
|
|
|
|
| 183 |
|
| 184 |
# Display the uploaded CSV data
|
| 185 |
st.write("Uploaded CSV Data:")
|
|
@@ -207,6 +236,12 @@ if csv_file is not None:
|
|
| 207 |
st.write(f"Question: {q}")
|
| 208 |
st.write(f"Answer: {a}")
|
| 209 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 210 |
# Add Ctrl+Enter functionality for submitting the questions
|
| 211 |
st.markdown("""
|
| 212 |
<script>
|
|
|
|
| 1 |
from copy import deepcopy
|
| 2 |
+
from langchain.callbacks import StreamlitCallbackHandler
|
| 3 |
import streamlit as st
|
| 4 |
import pandas as pd
|
| 5 |
from io import StringIO
|
|
|
|
| 9 |
from weaviate.embedded import EmbeddedOptions
|
| 10 |
from weaviate import Client
|
| 11 |
from weaviate.util import generate_uuid5
|
| 12 |
+
import logging
|
| 13 |
|
| 14 |
+
class StreamlitCallbackHandler(logging.Handler):
|
| 15 |
+
def emit(self, record):
|
| 16 |
+
log_entry = self.format(record)
|
| 17 |
+
st.write(log_entry)
|
| 18 |
+
|
| 19 |
# Initialize TAPAS model and tokenizer
|
| 20 |
tokenizer = AutoTokenizer.from_pretrained("google/tapas-large-finetuned-wtq")
|
| 21 |
model = AutoModelForTableQuestionAnswering.from_pretrained("google/tapas-large-finetuned-wtq")
|
|
|
|
| 25 |
embedded_options=EmbeddedOptions()
|
| 26 |
)
|
| 27 |
|
| 28 |
+
# Global list to store debugging information
|
| 29 |
+
DEBUG_LOGS = []
|
| 30 |
+
|
| 31 |
+
def log_debug_info(message):
|
| 32 |
+
if st.session_state.debug:
|
| 33 |
+
logger = logging.getLogger(__name__)
|
| 34 |
+
logger.setLevel(logging.DEBUG)
|
| 35 |
+
|
| 36 |
+
# Check if StreamlitCallbackHandler is already added to avoid duplicate logs
|
| 37 |
+
if not any(isinstance(handler, StreamlitCallbackHandler) for handler in logger.handlers):
|
| 38 |
+
handler = StreamlitCallbackHandler()
|
| 39 |
+
logger.addHandler(handler)
|
| 40 |
+
|
| 41 |
+
logger.debug(message)
|
| 42 |
+
|
| 43 |
+
|
| 44 |
# Function to check if a class already exists in Weaviate
|
| 45 |
def class_exists(class_name):
|
| 46 |
try:
|
|
|
|
| 99 |
}
|
| 100 |
client.data_object.create(obj)
|
| 101 |
|
| 102 |
+
# Log data ingestion
|
| 103 |
+
log_debug_info(f"Data ingested into Weaviate for class: {class_name}")
|
| 104 |
|
| 105 |
def query_weaviate(question):
|
| 106 |
# This is a basic example; adapt the query based on the question
|
|
|
|
| 112 |
try:
|
| 113 |
inputs = tokenizer(table=chunk, queries=questions, padding="max_length", truncation=True, return_tensors="pt")
|
| 114 |
except Exception as e:
|
| 115 |
+
log_debug_info(f"Tokenization error: {e}")
|
| 116 |
st.write(f"An error occurred: {e}")
|
| 117 |
return ["Error occurred while tokenizing"] * len(questions)
|
| 118 |
|
| 119 |
if inputs["input_ids"].shape[1] > 512:
|
| 120 |
+
log_debug_info("Token limit exceeded for chunk")
|
| 121 |
st.warning("Token limit exceeded for chunk")
|
| 122 |
return ["Token limit exceeded for chunk"] * len(questions)
|
| 123 |
|
|
|
|
| 133 |
if len(coordinates) == 1:
|
| 134 |
row, col = coordinates[0]
|
| 135 |
try:
|
|
|
|
|
|
|
|
|
|
| 136 |
value = chunk.iloc[row, col]
|
| 137 |
+
log_debug_info(f"Accessed value for row {row}, col {col}: {value}")
|
| 138 |
answers.append(value)
|
| 139 |
except Exception as e:
|
| 140 |
+
log_debug_info(f"Error accessing value for row {row}, col {col}: {e}")
|
| 141 |
st.write(f"An error occurred: {e}")
|
| 142 |
else:
|
| 143 |
cell_values = []
|
|
|
|
| 147 |
value = chunk.iloc[row, col]
|
| 148 |
cell_values.append(value)
|
| 149 |
except Exception as e:
|
| 150 |
+
log_debug_info(f"Error accessing value for row {row}, col {col}: {e}")
|
| 151 |
st.write(f"An error occurred: {e}")
|
| 152 |
answers.append(", ".join(map(str, cell_values)))
|
| 153 |
|
|
|
|
| 206 |
if csv_file is not None:
|
| 207 |
data = csv_file.read().decode("utf-8")
|
| 208 |
dataframe = pd.read_csv(StringIO(data))
|
| 209 |
+
|
| 210 |
+
# Log CSV upload information
|
| 211 |
+
log_debug_info(f"CSV uploaded with shape: {dataframe.shape}")
|
| 212 |
|
| 213 |
# Display the uploaded CSV data
|
| 214 |
st.write("Uploaded CSV Data:")
|
|
|
|
| 236 |
st.write(f"Question: {q}")
|
| 237 |
st.write(f"Answer: {a}")
|
| 238 |
|
| 239 |
+
# Display debugging information
|
| 240 |
+
if st.checkbox("Show Debugging Information"):
|
| 241 |
+
st.write("Debugging Logs:")
|
| 242 |
+
for log in DEBUG_LOGS:
|
| 243 |
+
st.write(log)
|
| 244 |
+
|
| 245 |
# Add Ctrl+Enter functionality for submitting the questions
|
| 246 |
st.markdown("""
|
| 247 |
<script>
|