Spaces:
Running
Running
App debug mode
#14
by AppleSwing - opened
app.py
CHANGED
|
@@ -3,10 +3,11 @@
|
|
| 3 |
import os
|
| 4 |
import datetime
|
| 5 |
import socket
|
|
|
|
| 6 |
|
| 7 |
import gradio as gr
|
| 8 |
import pandas as pd
|
| 9 |
-
|
| 10 |
from apscheduler.schedulers.background import BackgroundScheduler
|
| 11 |
|
| 12 |
from huggingface_hub import snapshot_download
|
|
@@ -38,11 +39,23 @@ from src.display.utils import (
|
|
| 38 |
Precision,
|
| 39 |
)
|
| 40 |
|
| 41 |
-
from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, H4_TOKEN, IS_PUBLIC, QUEUE_REPO, REPO_ID, RESULTS_REPO
|
| 42 |
from src.populate import get_evaluation_queue_df, get_leaderboard_df
|
| 43 |
from src.submission.submit import add_new_eval
|
| 44 |
from src.utils import get_dataset_summary_table
|
| 45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
|
| 47 |
def ui_snapshot_download(repo_id, local_dir, repo_type, tqdm_class, etag_timeout):
|
| 48 |
try:
|
|
@@ -154,7 +167,7 @@ demo = gr.Blocks(css=custom_css)
|
|
| 154 |
with demo:
|
| 155 |
gr.HTML(TITLE)
|
| 156 |
gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
|
| 157 |
-
|
| 158 |
with gr.Tabs(elem_classes="tab-buttons") as tabs:
|
| 159 |
with gr.TabItem("MOE-LLM-GPU-Poor-Leaderboard Benchmark", elem_id="llm-benchmark-tab-table", id=0):
|
| 160 |
with gr.Row():
|
|
@@ -233,7 +246,7 @@ with demo:
|
|
| 233 |
interactive=False,
|
| 234 |
visible=True,
|
| 235 |
) # column_widths=["2%", "20%"]
|
| 236 |
-
|
| 237 |
# Dummy leaderboard for handling the case when the user uses backspace key
|
| 238 |
hidden_leaderboard_table_for_search = gr.components.Dataframe(
|
| 239 |
value=original_df[COLS] if original_df.empty is False else original_df,
|
|
@@ -288,7 +301,7 @@ with demo:
|
|
| 288 |
|
| 289 |
gr.Markdown(LLM_BENCHMARKS_DETAILS, elem_classes="markdown-text")
|
| 290 |
gr.Markdown(FAQ_TEXT, elem_classes="markdown-text")
|
| 291 |
-
|
| 292 |
with gr.TabItem("Submit a model ", elem_id="llm-benchmark-tab-table", id=3):
|
| 293 |
with gr.Column():
|
| 294 |
with gr.Row():
|
|
@@ -373,7 +386,7 @@ with demo:
|
|
| 373 |
],
|
| 374 |
submission_result,
|
| 375 |
)
|
| 376 |
-
|
| 377 |
with gr.Row():
|
| 378 |
with gr.Accordion("Citing this leaderboard", open=False):
|
| 379 |
citation_button = gr.Textbox(
|
|
|
|
| 3 |
import os
|
| 4 |
import datetime
|
| 5 |
import socket
|
| 6 |
+
from threading import Thread
|
| 7 |
|
| 8 |
import gradio as gr
|
| 9 |
import pandas as pd
|
| 10 |
+
import time
|
| 11 |
from apscheduler.schedulers.background import BackgroundScheduler
|
| 12 |
|
| 13 |
from huggingface_hub import snapshot_download
|
|
|
|
| 39 |
Precision,
|
| 40 |
)
|
| 41 |
|
| 42 |
+
from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, H4_TOKEN, IS_PUBLIC, QUEUE_REPO, REPO_ID, RESULTS_REPO, DEBUG_QUEUE_REPO, DEBUG_RESULTS_REPO
|
| 43 |
from src.populate import get_evaluation_queue_df, get_leaderboard_df
|
| 44 |
from src.submission.submit import add_new_eval
|
| 45 |
from src.utils import get_dataset_summary_table
|
| 46 |
|
| 47 |
+
def get_args():
|
| 48 |
+
import argparse
|
| 49 |
+
|
| 50 |
+
parser = argparse.ArgumentParser(description="Run the LLM Leaderboard")
|
| 51 |
+
parser.add_argument("--debug", action="store_true", help="Run in debug mode")
|
| 52 |
+
return parser.parse_args()
|
| 53 |
+
|
| 54 |
+
args = get_args()
|
| 55 |
+
if args.debug:
|
| 56 |
+
print("Running in debug mode")
|
| 57 |
+
QUEUE_REPO = DEBUG_QUEUE_REPO
|
| 58 |
+
RESULTS_REPO = DEBUG_RESULTS_REPO
|
| 59 |
|
| 60 |
def ui_snapshot_download(repo_id, local_dir, repo_type, tqdm_class, etag_timeout):
|
| 61 |
try:
|
|
|
|
| 167 |
with demo:
|
| 168 |
gr.HTML(TITLE)
|
| 169 |
gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
|
| 170 |
+
|
| 171 |
with gr.Tabs(elem_classes="tab-buttons") as tabs:
|
| 172 |
with gr.TabItem("MOE-LLM-GPU-Poor-Leaderboard Benchmark", elem_id="llm-benchmark-tab-table", id=0):
|
| 173 |
with gr.Row():
|
|
|
|
| 246 |
interactive=False,
|
| 247 |
visible=True,
|
| 248 |
) # column_widths=["2%", "20%"]
|
| 249 |
+
|
| 250 |
# Dummy leaderboard for handling the case when the user uses backspace key
|
| 251 |
hidden_leaderboard_table_for_search = gr.components.Dataframe(
|
| 252 |
value=original_df[COLS] if original_df.empty is False else original_df,
|
|
|
|
| 301 |
|
| 302 |
gr.Markdown(LLM_BENCHMARKS_DETAILS, elem_classes="markdown-text")
|
| 303 |
gr.Markdown(FAQ_TEXT, elem_classes="markdown-text")
|
| 304 |
+
|
| 305 |
with gr.TabItem("Submit a model ", elem_id="llm-benchmark-tab-table", id=3):
|
| 306 |
with gr.Column():
|
| 307 |
with gr.Row():
|
|
|
|
| 386 |
],
|
| 387 |
submission_result,
|
| 388 |
)
|
| 389 |
+
|
| 390 |
with gr.Row():
|
| 391 |
with gr.Accordion("Citing this leaderboard", open=False):
|
| 392 |
citation_button = gr.Textbox(
|