Spaces:
Running
Running
Commit
·
820f694
1
Parent(s):
8a0d82d
update
Browse files
app.py
CHANGED
|
@@ -267,24 +267,7 @@ def run_evaluation(text: str, progress=gr.Progress()):
|
|
| 267 |
# Wrap HTML for iframe display
|
| 268 |
wrapped_html = wrap_html_in_iframe(html)
|
| 269 |
|
| 270 |
-
|
| 271 |
-
timing_lines = ["## ⏱️ Inference Timing\n"]
|
| 272 |
-
|
| 273 |
-
# Qwen timing
|
| 274 |
-
if qwen_predicted_time is not None:
|
| 275 |
-
timing_lines.append(f"**Qwen3-1.7B-Base**: {result_qwen['inference_time']:.2f}s (predicted: {qwen_predicted_time:.2f}s)")
|
| 276 |
-
else:
|
| 277 |
-
timing_lines.append(f"**Qwen3-1.7B-Base**: {result_qwen['inference_time']:.2f}s")
|
| 278 |
-
|
| 279 |
-
# RWKV timing
|
| 280 |
-
if rwkv_predicted_time is not None:
|
| 281 |
-
timing_lines.append(f"**RWKV7-G1C-1.5B**: {result_rwkv['inference_time']:.2f}s (predicted: {rwkv_predicted_time:.2f}s)")
|
| 282 |
-
else:
|
| 283 |
-
timing_lines.append(f"**RWKV7-G1C-1.5B**: {result_rwkv['inference_time']:.2f}s")
|
| 284 |
-
|
| 285 |
-
timing_text = "\n\n".join(timing_lines)
|
| 286 |
-
|
| 287 |
-
return wrapped_html, timing_text
|
| 288 |
|
| 289 |
except torch.cuda.OutOfMemoryError:
|
| 290 |
if torch.cuda.is_available():
|
|
@@ -300,7 +283,7 @@ def run_evaluation(text: str, progress=gr.Progress()):
|
|
| 300 |
|
| 301 |
def clear_inputs():
|
| 302 |
"""Clear all inputs and outputs."""
|
| 303 |
-
return "", None
|
| 304 |
|
| 305 |
|
| 306 |
def get_default_example():
|
|
@@ -309,9 +292,9 @@ def get_default_example():
|
|
| 309 |
|
| 310 |
if _precomputed_html and _precomputed_text:
|
| 311 |
wrapped_html = wrap_html_in_iframe(_precomputed_html)
|
| 312 |
-
return _precomputed_text, wrapped_html
|
| 313 |
else:
|
| 314 |
-
return "", None
|
| 315 |
|
| 316 |
|
| 317 |
# Build Gradio UI
|
|
@@ -338,22 +321,17 @@ with gr.Blocks(title="Compression-Lens: RWKV-7 vs Qwen3", theme=gr.themes.Soft()
|
|
| 338 |
|
| 339 |
gr.Markdown("---")
|
| 340 |
|
| 341 |
-
# Timing information display
|
| 342 |
-
with gr.Row():
|
| 343 |
-
with gr.Column():
|
| 344 |
-
timing_info = gr.Markdown(label="Inference Timing")
|
| 345 |
-
|
| 346 |
with gr.Row():
|
| 347 |
with gr.Column():
|
| 348 |
output_html = gr.HTML(label="Visualization")
|
| 349 |
|
| 350 |
# Event handlers
|
| 351 |
-
clear_btn.click(fn=clear_inputs, outputs=[text_input, output_html
|
| 352 |
|
| 353 |
-
run_btn.click(fn=run_evaluation, inputs=[text_input], outputs=[output_html
|
| 354 |
|
| 355 |
# Load default example on page load
|
| 356 |
-
demo.load(fn=get_default_example, outputs=[text_input, output_html
|
| 357 |
|
| 358 |
|
| 359 |
if __name__ == "__main__":
|
|
|
|
| 267 |
# Wrap HTML for iframe display
|
| 268 |
wrapped_html = wrap_html_in_iframe(html)
|
| 269 |
|
| 270 |
+
return wrapped_html
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 271 |
|
| 272 |
except torch.cuda.OutOfMemoryError:
|
| 273 |
if torch.cuda.is_available():
|
|
|
|
| 283 |
|
| 284 |
def clear_inputs():
|
| 285 |
"""Clear all inputs and outputs."""
|
| 286 |
+
return "", None
|
| 287 |
|
| 288 |
|
| 289 |
def get_default_example():
|
|
|
|
| 292 |
|
| 293 |
if _precomputed_html and _precomputed_text:
|
| 294 |
wrapped_html = wrap_html_in_iframe(_precomputed_html)
|
| 295 |
+
return _precomputed_text, wrapped_html
|
| 296 |
else:
|
| 297 |
+
return "", None
|
| 298 |
|
| 299 |
|
| 300 |
# Build Gradio UI
|
|
|
|
| 321 |
|
| 322 |
gr.Markdown("---")
|
| 323 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 324 |
with gr.Row():
|
| 325 |
with gr.Column():
|
| 326 |
output_html = gr.HTML(label="Visualization")
|
| 327 |
|
| 328 |
# Event handlers
|
| 329 |
+
clear_btn.click(fn=clear_inputs, outputs=[text_input, output_html])
|
| 330 |
|
| 331 |
+
run_btn.click(fn=run_evaluation, inputs=[text_input], outputs=[output_html])
|
| 332 |
|
| 333 |
# Load default example on page load
|
| 334 |
+
demo.load(fn=get_default_example, outputs=[text_input, output_html])
|
| 335 |
|
| 336 |
|
| 337 |
if __name__ == "__main__":
|