Spaces:
Runtime error
Runtime error
Commit
·
8bbf037
1
Parent(s):
76ed6d2
- __pycache__/main.cpython-310.pyc +0 -0
- main.py +46 -38
__pycache__/main.cpython-310.pyc
CHANGED
Binary files a/__pycache__/main.cpython-310.pyc and b/__pycache__/main.cpython-310.pyc differ
|
|
main.py
CHANGED
@@ -5,6 +5,7 @@ import time
|
|
5 |
from concurrent.futures import ThreadPoolExecutor, as_completed
|
6 |
import threading
|
7 |
import streamlit as st # Import Streamlit
|
|
|
8 |
|
9 |
|
10 |
def generate_answer(question, previous_answers, model_name, open_router_key, openai_api_key):
|
@@ -35,9 +36,8 @@ def evaluate_answer(question, new_answer, open_router_key, openai_api_key):
|
|
35 |
return None
|
36 |
|
37 |
|
38 |
-
def process_question(question, model_name, open_router_key, openai_api_key,
|
39 |
start_time = time.time()
|
40 |
-
st.write(f"<span style='color:red'>{question}</span>", unsafe_allow_html=True)
|
41 |
previous_answers = []
|
42 |
question_novelty = 0
|
43 |
|
@@ -52,39 +52,38 @@ def process_question(question, model_name, open_router_key, openai_api_key, prog
|
|
52 |
break
|
53 |
|
54 |
if coherence_score <= 3:
|
55 |
-
|
56 |
-
unsafe_allow_html=True)
|
57 |
break
|
58 |
|
59 |
novelty_score = get_novelty_score(new_answer, previous_answers, openai_api_key)
|
60 |
|
61 |
if novelty_score < 0.1:
|
62 |
-
st.write("<span style='color:yellow'>Output is redundant. Moving to next question.</span>",
|
63 |
-
unsafe_allow_html=True)
|
64 |
break
|
65 |
|
66 |
-
st.write
|
67 |
-
|
68 |
-
|
69 |
-
|
|
|
|
|
|
|
|
|
70 |
|
71 |
previous_answers.append(new_answer)
|
72 |
question_novelty += novelty_score
|
73 |
|
74 |
except Exception as e:
|
75 |
-
|
76 |
-
|
77 |
|
78 |
time_taken = time.time() - start_time
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
|
|
|
|
83 |
|
84 |
-
# Update progress
|
85 |
-
with progress_lock:
|
86 |
-
completed_questions += 1
|
87 |
-
progress = completed_questions / total_questions
|
88 |
|
89 |
return question_novelty, [
|
90 |
{
|
@@ -117,11 +116,10 @@ def get_novelty_score(new_answer: str, previous_answers: list, openai_api_key):
|
|
117 |
return novelty
|
118 |
|
119 |
|
120 |
-
def benchmark_model_multithreaded(model_name, questions, open_router_key, openai_api_key, max_threads=None
|
121 |
novelty_score = 0
|
122 |
-
print_lock = threading.Lock() # Lock for thread-safe printing
|
123 |
results = []
|
124 |
-
|
125 |
|
126 |
# Use max_threads if provided, otherwise default to the number of questions
|
127 |
if max_threads is None:
|
@@ -130,23 +128,33 @@ def benchmark_model_multithreaded(model_name, questions, open_router_key, openai
|
|
130 |
max_workers = max_threads
|
131 |
|
132 |
with ThreadPoolExecutor(max_workers=max_workers) as executor:
|
133 |
-
|
134 |
-
|
135 |
-
|
136 |
-
|
137 |
-
|
138 |
|
|
|
|
|
139 |
try:
|
140 |
-
|
141 |
-
|
142 |
-
|
143 |
-
|
144 |
-
st.write(
|
145 |
-
|
146 |
-
|
147 |
-
|
148 |
-
|
149 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
150 |
|
151 |
st.write(f"<span style='color:yellow'>Final total novelty score across all questions: {novelty_score}</span>",
|
152 |
unsafe_allow_html=True)
|
|
|
5 |
from concurrent.futures import ThreadPoolExecutor, as_completed
|
6 |
import threading
|
7 |
import streamlit as st # Import Streamlit
|
8 |
+
import queue
|
9 |
|
10 |
|
11 |
def generate_answer(question, previous_answers, model_name, open_router_key, openai_api_key):
|
|
|
36 |
return None
|
37 |
|
38 |
|
39 |
+
def process_question(question, model_name, open_router_key, openai_api_key, result_queue):
|
40 |
start_time = time.time()
|
|
|
41 |
previous_answers = []
|
42 |
question_novelty = 0
|
43 |
|
|
|
52 |
break
|
53 |
|
54 |
if coherence_score <= 3:
|
55 |
+
|
|
|
56 |
break
|
57 |
|
58 |
novelty_score = get_novelty_score(new_answer, previous_answers, openai_api_key)
|
59 |
|
60 |
if novelty_score < 0.1:
|
|
|
|
|
61 |
break
|
62 |
|
63 |
+
# Append results to the queue instead of using st.write
|
64 |
+
result_queue.put({
|
65 |
+
"type": "answer",
|
66 |
+
"question": question,
|
67 |
+
"answer": new_answer,
|
68 |
+
"coherence_score": coherence_score,
|
69 |
+
"novelty_score": novelty_score
|
70 |
+
})
|
71 |
|
72 |
previous_answers.append(new_answer)
|
73 |
question_novelty += novelty_score
|
74 |
|
75 |
except Exception as e:
|
76 |
+
result_queue.put({"type": "error", "message": str(e)})
|
77 |
+
|
78 |
|
79 |
time_taken = time.time() - start_time
|
80 |
+
result_queue.put({
|
81 |
+
"type": "summary",
|
82 |
+
"question": question,
|
83 |
+
"total_novelty": question_novelty,
|
84 |
+
"time_taken": time_taken
|
85 |
+
})
|
86 |
|
|
|
|
|
|
|
|
|
87 |
|
88 |
return question_novelty, [
|
89 |
{
|
|
|
116 |
return novelty
|
117 |
|
118 |
|
119 |
+
def benchmark_model_multithreaded(model_name, questions, open_router_key, openai_api_key, max_threads=None):
|
120 |
novelty_score = 0
|
|
|
121 |
results = []
|
122 |
+
result_queue = queue.Queue() # Create a queue for communication
|
123 |
|
124 |
# Use max_threads if provided, otherwise default to the number of questions
|
125 |
if max_threads is None:
|
|
|
128 |
max_workers = max_threads
|
129 |
|
130 |
with ThreadPoolExecutor(max_workers=max_workers) as executor:
|
131 |
+
# Submit tasks to the thread pool
|
132 |
+
future_to_question = {
|
133 |
+
executor.submit(process_question, question, model_name, open_router_key, openai_api_key, result_queue): question
|
134 |
+
for question in questions
|
135 |
+
}
|
136 |
|
137 |
+
# Process results from the queue in the main thread
|
138 |
+
while True:
|
139 |
try:
|
140 |
+
result = result_queue.get_nowait()
|
141 |
+
if result["type"] == "answer":
|
142 |
+
st.write(f"**Question:** {result['question']}")
|
143 |
+
st.write(f"**New Answer:**\n{result['answer']}")
|
144 |
+
st.write(f"<span style='color:green'>Coherence Score: {result['coherence_score']}</span>",
|
145 |
+
unsafe_allow_html=True)
|
146 |
+
st.write(f"**Novelty Score:** {result['novelty_score']}")
|
147 |
+
elif result["type"] == "summary":
|
148 |
+
st.write(f"<span style='color:blue'>Total novelty score for question '{result['question']}': {result['total_novelty']}</span>",
|
149 |
+
unsafe_allow_html=True)
|
150 |
+
st.write(f"<span style='color:blue'>Time taken: {result['time_taken']} seconds</span>",
|
151 |
+
unsafe_allow_html=True)
|
152 |
+
elif result["type"] == "error":
|
153 |
+
st.write(f"<span style='color:red'>Error in thread: {result['message']}</span>",
|
154 |
+
unsafe_allow_html=True)
|
155 |
+
except queue.Empty:
|
156 |
+
if not any(future.running() for future in future_to_question.keys()):
|
157 |
+
break # All tasks are done
|
158 |
|
159 |
st.write(f"<span style='color:yellow'>Final total novelty score across all questions: {novelty_score}</span>",
|
160 |
unsafe_allow_html=True)
|