File size: 6,940 Bytes
ab59957 43ee4de 1c73b10 3ce2f84 43ee4de 1c73b10 ab59957 1c73b10 ab59957 1c73b10 3ce2f84 1c73b10 a7fa922 1c73b10 3ce2f84 1c73b10 3ce2f84 1c73b10 3b81b14 1c73b10 3ce2f84 1c73b10 43ee4de 1c73b10 a7fa922 1c73b10 3b81b14 1c73b10 a7fa922 1c73b10 3ce2f84 a7fa922 3ce2f84 a7fa922 1c73b10 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 |
import gradio as gr
from apscheduler.schedulers.background import BackgroundScheduler
from typing import Optional
import logging
import sys
import time
from config import CONFIG
from data_manager import data_manager
from utils import filter_leaderboard, search_responses, plot_section_results, validate_model_submission
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
def create_app() -> gr.Blocks:
"""Create and configure the Gradio application."""
# Pre-load data with retries to avoid startup failures
def safe_get_data():
max_attempts = 3
for attempt in range(max_attempts):
try:
logger.info(f"Pre-loading data (attempt {attempt+1}/{max_attempts})...")
# Try to access data to trigger loading
families = data_manager.leaderboard_data["family"].unique().tolist() if not data_manager.leaderboard_data.empty else []
models = data_manager.leaderboard_data["model"].unique().tolist() if not data_manager.leaderboard_data.empty else []
logger.info(f"Successfully loaded data with {len(families)} families and {len(models)} models")
return True
except Exception as e:
logger.error(f"Error pre-loading data: {e}")
if attempt < max_attempts - 1:
logger.info(f"Retrying in {CONFIG['dataset'].retry_delay} seconds...")
time.sleep(CONFIG["dataset"].retry_delay)
else:
logger.warning("Using fallback data due to loading failures")
return False
# Try to pre-load data
safe_get_data()
with gr.Blocks(css=CONFIG["ui"].css, theme=CONFIG["ui"].theme) as app:
gr.HTML(f"<h1>{CONFIG['ui'].title}</h1>")
gr.Markdown(CONFIG["ui"].description)
with gr.Tabs() as tabs:
# Leaderboard Tab
with gr.TabItem("📊 Leaderboard"):
with gr.Row():
family_filter = gr.Dropdown(
choices=data_manager.leaderboard_data["family"].unique().tolist() if not data_manager.leaderboard_data.empty else [],
label="Filter by Family",
multiselect=False
)
quantization_filter = gr.Dropdown(
choices=data_manager.leaderboard_data["quantization_level"].unique().tolist() if not data_manager.leaderboard_data.empty else [],
label="Filter by Quantization Level"
)
filter_btn = gr.Button("Apply Filters", variant="primary")
leaderboard_table = gr.DataFrame(
value=data_manager.leaderboard_data,
interactive=False
)
filter_btn.click(
filter_leaderboard,
inputs=[family_filter, quantization_filter],
outputs=leaderboard_table
)
# Model Responses Tab
with gr.TabItem("🔍 Model Responses"):
with gr.Row():
model_dropdown = gr.Dropdown(
choices=data_manager.leaderboard_data["model"].unique().tolist() if not data_manager.leaderboard_data.empty else [],
label="Select Model"
)
query_input = gr.Textbox(
label="Search Query",
placeholder="Enter search terms..."
)
search_btn = gr.Button("Search", variant="primary")
responses_table = gr.DataFrame()
search_btn.click(
search_responses,
inputs=[query_input, model_dropdown],
outputs=responses_table
)
# Section Results Tab
with gr.TabItem("📈 Section Results"):
gr.Plot(value=plot_section_results)
gr.DataFrame(value=data_manager.section_results_data)
# Submit Model Tab
with gr.TabItem("➕ Submit Model"):
gr.Markdown("### Submit Your Model for Evaluation")
with gr.Group():
model_name = gr.Textbox(label="Model Name", placeholder="Enter unique model name")
base_model = gr.Textbox(label="Base Model", placeholder="Enter base model name")
revision = gr.Textbox(label="Revision", value="main")
with gr.Row():
precision = gr.Dropdown(
choices=CONFIG["model"].precision_options,
label="Precision",
value="float16"
)
weight_type = gr.Dropdown(
choices=CONFIG["model"].weight_types,
label="Weight Type",
value="Original"
)
model_type = gr.Dropdown(
choices=CONFIG["model"].model_types,
label="Model Type",
value="Transformer"
)
submit_btn = gr.Button("Submit Model", variant="primary")
submission_output = gr.Markdown()
def handle_submission(*args):
is_valid, message = validate_model_submission(*args)
if not is_valid:
return f"❌ {message}"
return "✅ Model submitted successfully!"
submit_btn.click(
handle_submission,
inputs=[model_name, base_model, revision, precision, weight_type, model_type],
outputs=submission_output
)
return app
def main():
try:
# Initialize scheduler for data refresh
scheduler = BackgroundScheduler()
scheduler.add_job(
data_manager.refresh_datasets,
"interval",
seconds=CONFIG["dataset"].refresh_interval
)
scheduler.start()
# Create and launch app
app = create_app()
app.queue(default_concurrency_limit=40).launch(
inbrowser=True,
server_name="0.0.0.0", # Use 0.0.0.0 to listen on all interfaces
server_port=7860,
share=False,
debug=False,
show_error=True,
max_threads=40
)
except Exception as e:
logger.error(f"Error starting application: {e}")
sys.exit(1)
if __name__ == "__main__":
main() |