Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -27,7 +27,7 @@ class AIEvaluationForm:
|
|
| 27 |
components = {}
|
| 28 |
|
| 29 |
with gr.Group():
|
| 30 |
-
gr.Markdown("##
|
| 31 |
gr.Markdown("*Please provide basic information about the AI system being evaluated.*")
|
| 32 |
|
| 33 |
components['name'] = gr.Textbox(
|
|
@@ -61,7 +61,7 @@ class AIEvaluationForm:
|
|
| 61 |
info="Primary category of the AI system"
|
| 62 |
)
|
| 63 |
|
| 64 |
-
components['
|
| 65 |
choices=[
|
| 66 |
"Text",
|
| 67 |
"Image",
|
|
@@ -71,10 +71,10 @@ class AIEvaluationForm:
|
|
| 71 |
],
|
| 72 |
label="Input modalities (select all that apply)",
|
| 73 |
value=["Text"],
|
| 74 |
-
info="
|
| 75 |
)
|
| 76 |
|
| 77 |
-
components['
|
| 78 |
choices=[
|
| 79 |
"Text",
|
| 80 |
"Image",
|
|
@@ -84,7 +84,7 @@ class AIEvaluationForm:
|
|
| 84 |
],
|
| 85 |
label="Output Modalities (select all that apply)",
|
| 86 |
value=["Text"],
|
| 87 |
-
info="
|
| 88 |
)
|
| 89 |
|
| 90 |
return list(components.values()), components
|
|
@@ -163,13 +163,13 @@ class AIEvaluationForm:
|
|
| 163 |
|
| 164 |
# Determine source type based on content
|
| 165 |
if line.startswith('http'):
|
| 166 |
-
source_type = "
|
| 167 |
name = line.split('/')[-1] if '/' in line else line
|
| 168 |
elif 'internal' in line.lower() or 'proprietary' in line.lower():
|
| 169 |
-
source_type = "
|
| 170 |
name = line
|
| 171 |
else:
|
| 172 |
-
source_type = "
|
| 173 |
name = line
|
| 174 |
|
| 175 |
sources.append({
|
|
@@ -179,6 +179,17 @@ class AIEvaluationForm:
|
|
| 179 |
})
|
| 180 |
|
| 181 |
return sources
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 182 |
|
| 183 |
def generate_scorecard(self, *args) -> Tuple[Dict, str]:
|
| 184 |
"""Generate scorecard JSON from form inputs"""
|
|
@@ -187,10 +198,9 @@ class AIEvaluationForm:
|
|
| 187 |
for i, arg in enumerate(args[:10]): # Print first 10 for debugging
|
| 188 |
print(f"Arg {i}: {type(arg)} = {arg}")
|
| 189 |
|
| 190 |
-
# Extract system info (first
|
| 191 |
-
|
| 192 |
-
|
| 193 |
-
remaining_args = list(args[num_args:])
|
| 194 |
|
| 195 |
# Build metadata
|
| 196 |
metadata = {
|
|
@@ -308,7 +318,7 @@ class AIEvaluationForm:
|
|
| 308 |
|
| 309 |
# Header
|
| 310 |
gr.Markdown("""
|
| 311 |
-
#
|
| 312 |
|
| 313 |
This comprehensive evaluation form helps you assess AI systems across multiple dimensions including bias,
|
| 314 |
cultural sensitivity, environmental impact, privacy, and more. Complete the sections relevant to your system
|
|
@@ -326,17 +336,17 @@ class AIEvaluationForm:
|
|
| 326 |
|
| 327 |
# Generate button and outputs
|
| 328 |
with gr.Group():
|
| 329 |
-
gr.Markdown("##
|
| 330 |
|
| 331 |
with gr.Row():
|
| 332 |
generate_btn = gr.Button(
|
| 333 |
-
"
|
| 334 |
variant="primary",
|
| 335 |
size="lg",
|
| 336 |
scale=2
|
| 337 |
)
|
| 338 |
clear_btn = gr.Button(
|
| 339 |
-
"
|
| 340 |
variant="secondary",
|
| 341 |
scale=1
|
| 342 |
)
|
|
@@ -346,7 +356,7 @@ class AIEvaluationForm:
|
|
| 346 |
|
| 347 |
# Outputs
|
| 348 |
with gr.Group():
|
| 349 |
-
gr.Markdown("###
|
| 350 |
|
| 351 |
with gr.Row():
|
| 352 |
json_output = gr.JSON(
|
|
@@ -359,13 +369,6 @@ class AIEvaluationForm:
|
|
| 359 |
label="Download Scorecard",
|
| 360 |
visible=False
|
| 361 |
)
|
| 362 |
-
download_btn = gr.Button(
|
| 363 |
-
"πΎ Download JSON",
|
| 364 |
-
visible=False,
|
| 365 |
-
variant="secondary"
|
| 366 |
-
)
|
| 367 |
-
# Hidden HTML component for triggering downloads
|
| 368 |
-
download_html = gr.HTML(visible=False)
|
| 369 |
|
| 370 |
# Event handlers
|
| 371 |
all_inputs = system_inputs + eval_inputs
|
|
@@ -386,25 +389,9 @@ class AIEvaluationForm:
|
|
| 386 |
|
| 387 |
return (
|
| 388 |
scorecard, # JSON display
|
| 389 |
-
gr.File(value=filename, visible=True)
|
| 390 |
-
gr.Button(visible=True), # Show download button
|
| 391 |
-
gr.HTML(visible=False) # Hidden HTML for download trigger
|
| 392 |
)
|
| 393 |
-
|
| 394 |
-
def trigger_download():
|
| 395 |
-
"""Trigger file download via JavaScript"""
|
| 396 |
-
return gr.HTML("""
|
| 397 |
-
<script>
|
| 398 |
-
// Find the download link and click it
|
| 399 |
-
setTimeout(() => {
|
| 400 |
-
const downloadLinks = document.querySelectorAll('a[download]');
|
| 401 |
-
if (downloadLinks.length > 0) {
|
| 402 |
-
downloadLinks[downloadLinks.length - 1].click();
|
| 403 |
-
}
|
| 404 |
-
}, 100);
|
| 405 |
-
</script>
|
| 406 |
-
""", visible=False)
|
| 407 |
-
|
| 408 |
def clear_form():
|
| 409 |
"""Clear all form inputs"""
|
| 410 |
return [None] * len(all_inputs)
|
|
@@ -413,16 +400,10 @@ class AIEvaluationForm:
|
|
| 413 |
generate_btn.click(
|
| 414 |
fn=generate_with_progress,
|
| 415 |
inputs=all_inputs,
|
| 416 |
-
outputs=[json_output, download_file
|
| 417 |
show_progress="full"
|
| 418 |
)
|
| 419 |
|
| 420 |
-
# Wire up download button to trigger download
|
| 421 |
-
download_btn.click(
|
| 422 |
-
fn=trigger_download,
|
| 423 |
-
outputs=download_html
|
| 424 |
-
)
|
| 425 |
-
|
| 426 |
clear_btn.click(
|
| 427 |
fn=clear_form,
|
| 428 |
outputs=all_inputs
|
|
@@ -430,8 +411,8 @@ class AIEvaluationForm:
|
|
| 430 |
|
| 431 |
# Add example data button
|
| 432 |
with gr.Group():
|
| 433 |
-
gr.Markdown("###
|
| 434 |
-
example_btn = gr.Button("
|
| 435 |
|
| 436 |
def load_example():
|
| 437 |
"""Load example data for StarCoder2-like system"""
|
|
@@ -440,8 +421,8 @@ class AIEvaluationForm:
|
|
| 440 |
"BigCode", # provider
|
| 441 |
"https://huggingface.co/bigcode/starcoder2-15b", # url
|
| 442 |
"Generative Model", # type
|
| 443 |
-
["Text"], # input modalities
|
| 444 |
-
["Text"]
|
| 445 |
]
|
| 446 |
# Add default values for evaluation sections (all N/A initially)
|
| 447 |
remaining_defaults = []
|
|
@@ -459,28 +440,21 @@ class AIEvaluationForm:
|
|
| 459 |
fn=load_example,
|
| 460 |
outputs=all_inputs
|
| 461 |
)
|
|
|
|
| 462 |
with gr.Group():
|
| 463 |
-
gr.Markdown("###
|
| 464 |
uploaded_file = gr.File(label="Upload JSON File", file_types=[".json"])
|
| 465 |
uploaded_preview = gr.JSON(label="Preview of Uploaded Content")
|
| 466 |
-
uploaded_file.change(fn=load_uploaded_json, inputs=uploaded_file, outputs=uploaded_preview)
|
| 467 |
|
| 468 |
gr.Markdown("""
|
| 469 |
-
###
|
| 470 |
Once downloaded, you can contribute by submitting a pull request to [Eval Cards GitHub](https://github.com/evaleval/Eval_Cards).
|
| 471 |
Place your file in the `submissions/` directory.
|
| 472 |
""")
|
| 473 |
|
| 474 |
return demo
|
| 475 |
-
|
| 476 |
-
def load_uploaded_json(file):
|
| 477 |
-
if file is None:
|
| 478 |
-
return {}
|
| 479 |
-
try:
|
| 480 |
-
with open(file.name, 'r') as f:
|
| 481 |
-
return json.load(f)
|
| 482 |
-
except Exception as e:
|
| 483 |
-
return {"error": str(e)}
|
| 484 |
|
| 485 |
def main():
|
| 486 |
"""Main function to run the application"""
|
|
@@ -491,9 +465,9 @@ def main():
|
|
| 491 |
# Create and launch the interface
|
| 492 |
demo = eval_form.create_interface()
|
| 493 |
|
| 494 |
-
print("
|
| 495 |
-
print(f"
|
| 496 |
-
print(f"
|
| 497 |
|
| 498 |
# Count total questions
|
| 499 |
total_questions = sum(
|
|
@@ -501,7 +475,7 @@ def main():
|
|
| 501 |
for section in eval_form.template.values()
|
| 502 |
for subsection in section.values()
|
| 503 |
)
|
| 504 |
-
print(f"
|
| 505 |
|
| 506 |
demo.launch(
|
| 507 |
ssr_mode=False,
|
|
@@ -512,10 +486,10 @@ def main():
|
|
| 512 |
)
|
| 513 |
|
| 514 |
except FileNotFoundError as e:
|
| 515 |
-
print(f"
|
| 516 |
print("Please ensure 'questions.yaml' exists in the current directory.")
|
| 517 |
except Exception as e:
|
| 518 |
-
print(f"
|
| 519 |
|
| 520 |
if __name__ == "__main__":
|
| 521 |
main()
|
|
|
|
| 27 |
components = {}
|
| 28 |
|
| 29 |
with gr.Group():
|
| 30 |
+
gr.Markdown("## System Information")
|
| 31 |
gr.Markdown("*Please provide basic information about the AI system being evaluated.*")
|
| 32 |
|
| 33 |
components['name'] = gr.Textbox(
|
|
|
|
| 61 |
info="Primary category of the AI system"
|
| 62 |
)
|
| 63 |
|
| 64 |
+
components['input_modalities'] = gr.CheckboxGroup(
|
| 65 |
choices=[
|
| 66 |
"Text",
|
| 67 |
"Image",
|
|
|
|
| 71 |
],
|
| 72 |
label="Input modalities (select all that apply)",
|
| 73 |
value=["Text"],
|
| 74 |
+
info="Input modalities supported by the system"
|
| 75 |
)
|
| 76 |
|
| 77 |
+
components['output_modalities'] = gr.CheckboxGroup(
|
| 78 |
choices=[
|
| 79 |
"Text",
|
| 80 |
"Image",
|
|
|
|
| 84 |
],
|
| 85 |
label="Output Modalities (select all that apply)",
|
| 86 |
value=["Text"],
|
| 87 |
+
info="Output modalities supported by the system"
|
| 88 |
)
|
| 89 |
|
| 90 |
return list(components.values()), components
|
|
|
|
| 163 |
|
| 164 |
# Determine source type based on content
|
| 165 |
if line.startswith('http'):
|
| 166 |
+
source_type = "Web"
|
| 167 |
name = line.split('/')[-1] if '/' in line else line
|
| 168 |
elif 'internal' in line.lower() or 'proprietary' in line.lower():
|
| 169 |
+
source_type = "Internal"
|
| 170 |
name = line
|
| 171 |
else:
|
| 172 |
+
source_type = "Document"
|
| 173 |
name = line
|
| 174 |
|
| 175 |
sources.append({
|
|
|
|
| 179 |
})
|
| 180 |
|
| 181 |
return sources
|
| 182 |
+
|
| 183 |
+
def load_uploaded_json(self, file):
|
| 184 |
+
"""Load JSON from uploaded file"""
|
| 185 |
+
if file is None:
|
| 186 |
+
return {}
|
| 187 |
+
try:
|
| 188 |
+
with open(file.name, 'r') as f:
|
| 189 |
+
return json.load(f)
|
| 190 |
+
except Exception as e:
|
| 191 |
+
return {"error": str(e)}
|
| 192 |
+
|
| 193 |
|
| 194 |
def generate_scorecard(self, *args) -> Tuple[Dict, str]:
|
| 195 |
"""Generate scorecard JSON from form inputs"""
|
|
|
|
| 198 |
for i, arg in enumerate(args[:10]): # Print first 10 for debugging
|
| 199 |
print(f"Arg {i}: {type(arg)} = {arg}")
|
| 200 |
|
| 201 |
+
# Extract system info (first 6 arguments)
|
| 202 |
+
name, provider, url, sys_type, inp_modalities, out_modalities = args[:6]
|
| 203 |
+
remaining_args = list(args[6:])
|
|
|
|
| 204 |
|
| 205 |
# Build metadata
|
| 206 |
metadata = {
|
|
|
|
| 318 |
|
| 319 |
# Header
|
| 320 |
gr.Markdown("""
|
| 321 |
+
# AI System Evaluation Scorecard
|
| 322 |
|
| 323 |
This comprehensive evaluation form helps you assess AI systems across multiple dimensions including bias,
|
| 324 |
cultural sensitivity, environmental impact, privacy, and more. Complete the sections relevant to your system
|
|
|
|
| 336 |
|
| 337 |
# Generate button and outputs
|
| 338 |
with gr.Group():
|
| 339 |
+
gr.Markdown("## Generate Scorecard")
|
| 340 |
|
| 341 |
with gr.Row():
|
| 342 |
generate_btn = gr.Button(
|
| 343 |
+
"Generate Evaluation Scorecard",
|
| 344 |
variant="primary",
|
| 345 |
size="lg",
|
| 346 |
scale=2
|
| 347 |
)
|
| 348 |
clear_btn = gr.Button(
|
| 349 |
+
"Clear Form",
|
| 350 |
variant="secondary",
|
| 351 |
scale=1
|
| 352 |
)
|
|
|
|
| 356 |
|
| 357 |
# Outputs
|
| 358 |
with gr.Group():
|
| 359 |
+
gr.Markdown("### Generated Scorecard")
|
| 360 |
|
| 361 |
with gr.Row():
|
| 362 |
json_output = gr.JSON(
|
|
|
|
| 369 |
label="Download Scorecard",
|
| 370 |
visible=False
|
| 371 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 372 |
|
| 373 |
# Event handlers
|
| 374 |
all_inputs = system_inputs + eval_inputs
|
|
|
|
| 389 |
|
| 390 |
return (
|
| 391 |
scorecard, # JSON display
|
| 392 |
+
gr.File(value=filename, visible=True) # File for download
|
|
|
|
|
|
|
| 393 |
)
|
| 394 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 395 |
def clear_form():
|
| 396 |
"""Clear all form inputs"""
|
| 397 |
return [None] * len(all_inputs)
|
|
|
|
| 400 |
generate_btn.click(
|
| 401 |
fn=generate_with_progress,
|
| 402 |
inputs=all_inputs,
|
| 403 |
+
outputs=[json_output, download_file],
|
| 404 |
show_progress="full"
|
| 405 |
)
|
| 406 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 407 |
clear_btn.click(
|
| 408 |
fn=clear_form,
|
| 409 |
outputs=all_inputs
|
|
|
|
| 411 |
|
| 412 |
# Add example data button
|
| 413 |
with gr.Group():
|
| 414 |
+
gr.Markdown("### Quick Start")
|
| 415 |
+
example_btn = gr.Button("Load Example Data", variant="secondary")
|
| 416 |
|
| 417 |
def load_example():
|
| 418 |
"""Load example data for StarCoder2-like system"""
|
|
|
|
| 421 |
"BigCode", # provider
|
| 422 |
"https://huggingface.co/bigcode/starcoder2-15b", # url
|
| 423 |
"Generative Model", # type
|
| 424 |
+
["Text"], # input modalities - FIXED: Added missing comma
|
| 425 |
+
["Text"] # output modalities
|
| 426 |
]
|
| 427 |
# Add default values for evaluation sections (all N/A initially)
|
| 428 |
remaining_defaults = []
|
|
|
|
| 440 |
fn=load_example,
|
| 441 |
outputs=all_inputs
|
| 442 |
)
|
| 443 |
+
|
| 444 |
with gr.Group():
|
| 445 |
+
gr.Markdown("### Upload Completed Evaluation JSON")
|
| 446 |
uploaded_file = gr.File(label="Upload JSON File", file_types=[".json"])
|
| 447 |
uploaded_preview = gr.JSON(label="Preview of Uploaded Content")
|
| 448 |
+
uploaded_file.change(fn=self.load_uploaded_json, inputs=uploaded_file, outputs=uploaded_preview)
|
| 449 |
|
| 450 |
gr.Markdown("""
|
| 451 |
+
### Submit Your Scorecard to the Eval Cards Repository
|
| 452 |
Once downloaded, you can contribute by submitting a pull request to [Eval Cards GitHub](https://github.com/evaleval/Eval_Cards).
|
| 453 |
Place your file in the `submissions/` directory.
|
| 454 |
""")
|
| 455 |
|
| 456 |
return demo
|
| 457 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 458 |
|
| 459 |
def main():
|
| 460 |
"""Main function to run the application"""
|
|
|
|
| 465 |
# Create and launch the interface
|
| 466 |
demo = eval_form.create_interface()
|
| 467 |
|
| 468 |
+
print("Launching AI Evaluation Scorecard...")
|
| 469 |
+
print(f"Loading questions from: {eval_form.template_file}")
|
| 470 |
+
print(f"Found {len(eval_form.template)} evaluation categories")
|
| 471 |
|
| 472 |
# Count total questions
|
| 473 |
total_questions = sum(
|
|
|
|
| 475 |
for section in eval_form.template.values()
|
| 476 |
for subsection in section.values()
|
| 477 |
)
|
| 478 |
+
print(f"Total evaluation questions: {total_questions}")
|
| 479 |
|
| 480 |
demo.launch(
|
| 481 |
ssr_mode=False,
|
|
|
|
| 486 |
)
|
| 487 |
|
| 488 |
except FileNotFoundError as e:
|
| 489 |
+
print(f"Error: {e}")
|
| 490 |
print("Please ensure 'questions.yaml' exists in the current directory.")
|
| 491 |
except Exception as e:
|
| 492 |
+
print(f"Unexpected error: {e}")
|
| 493 |
|
| 494 |
if __name__ == "__main__":
|
| 495 |
main()
|