Spaces:
Build error
Build error
prepare the app for just testing we will submit later
Browse files
app.py
CHANGED
|
@@ -96,53 +96,55 @@ def run_and_submit_all( profile: gr.OAuthProfile | None):
|
|
| 96 |
status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
|
| 97 |
print(status_update)
|
| 98 |
|
| 99 |
-
|
| 100 |
-
|
| 101 |
-
|
| 102 |
-
response = requests.post(submit_url, json=submission_data, timeout=60)
|
| 103 |
-
response.raise_for_status()
|
| 104 |
-
result_data = response.json()
|
| 105 |
-
final_status = (
|
| 106 |
-
f"Submission Successful!\n"
|
| 107 |
-
f"User: {result_data.get('username')}\n"
|
| 108 |
-
f"Overall Score: {result_data.get('score', 'N/A')}% "
|
| 109 |
-
f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
|
| 110 |
-
f"Message: {result_data.get('message', 'No message received.')}"
|
| 111 |
-
)
|
| 112 |
-
print("Submission successful.")
|
| 113 |
-
results_df = pd.DataFrame(results_log)
|
| 114 |
-
return final_status, results_df
|
| 115 |
-
except requests.exceptions.HTTPError as e:
|
| 116 |
-
error_detail = f"Server responded with status {e.response.status_code}."
|
| 117 |
try:
|
| 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 |
# --- Build Gradio Interface using Blocks ---
|
| 144 |
with gr.Blocks() as demo:
|
| 145 |
-
gr.Markdown("# Basic Agent Evaluation Runner")
|
| 146 |
gr.Markdown(
|
| 147 |
"""
|
| 148 |
**Instructions:**
|
|
@@ -160,7 +162,7 @@ with gr.Blocks() as demo:
|
|
| 160 |
|
| 161 |
gr.LoginButton()
|
| 162 |
|
| 163 |
-
run_button = gr.Button("Run Evaluation & Submit All Answers")
|
| 164 |
|
| 165 |
status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
|
| 166 |
# Removed max_rows=10 from DataFrame constructor
|
|
|
|
| 96 |
status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
|
| 97 |
print(status_update)
|
| 98 |
|
| 99 |
+
if False:
|
| 100 |
+
# 5. Submit
|
| 101 |
+
print(f"Submitting {len(answers_payload)} answers to: {submit_url}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
try:
|
| 103 |
+
response = requests.post(submit_url, json=submission_data, timeout=60)
|
| 104 |
+
response.raise_for_status()
|
| 105 |
+
result_data = response.json()
|
| 106 |
+
final_status = (
|
| 107 |
+
f"Submission Successful!\n"
|
| 108 |
+
f"User: {result_data.get('username')}\n"
|
| 109 |
+
f"Overall Score: {result_data.get('score', 'N/A')}% "
|
| 110 |
+
f"({result_data.get('correct_count', '?')}/{result_data.get('total_attempted', '?')} correct)\n"
|
| 111 |
+
f"Message: {result_data.get('message', 'No message received.')}"
|
| 112 |
+
)
|
| 113 |
+
print("Submission successful.")
|
| 114 |
+
results_df = pd.DataFrame(results_log)
|
| 115 |
+
return final_status, results_df
|
| 116 |
+
except requests.exceptions.HTTPError as e:
|
| 117 |
+
error_detail = f"Server responded with status {e.response.status_code}."
|
| 118 |
+
try:
|
| 119 |
+
error_json = e.response.json()
|
| 120 |
+
error_detail += f" Detail: {error_json.get('detail', e.response.text)}"
|
| 121 |
+
except requests.exceptions.JSONDecodeError:
|
| 122 |
+
error_detail += f" Response: {e.response.text[:500]}"
|
| 123 |
+
status_message = f"Submission Failed: {error_detail}"
|
| 124 |
+
print(status_message)
|
| 125 |
+
results_df = pd.DataFrame(results_log)
|
| 126 |
+
return status_message, results_df
|
| 127 |
+
except requests.exceptions.Timeout:
|
| 128 |
+
status_message = "Submission Failed: The request timed out."
|
| 129 |
+
print(status_message)
|
| 130 |
+
results_df = pd.DataFrame(results_log)
|
| 131 |
+
return status_message, results_df
|
| 132 |
+
except requests.exceptions.RequestException as e:
|
| 133 |
+
status_message = f"Submission Failed: Network error - {e}"
|
| 134 |
+
print(status_message)
|
| 135 |
+
results_df = pd.DataFrame(results_log)
|
| 136 |
+
return status_message, results_df
|
| 137 |
+
except Exception as e:
|
| 138 |
+
status_message = f"An unexpected error occurred during submission: {e}"
|
| 139 |
+
print(status_message)
|
| 140 |
+
results_df = pd.DataFrame(results_log)
|
| 141 |
+
return status_message, results_df
|
| 142 |
+
|
| 143 |
|
| 144 |
|
| 145 |
# --- Build Gradio Interface using Blocks ---
|
| 146 |
with gr.Blocks() as demo:
|
| 147 |
+
gr.Markdown("# Walid's Basic Agent Evaluation Runner")
|
| 148 |
gr.Markdown(
|
| 149 |
"""
|
| 150 |
**Instructions:**
|
|
|
|
| 162 |
|
| 163 |
gr.LoginButton()
|
| 164 |
|
| 165 |
+
run_button = gr.Button("Run Evaluation (& Submit All Answers)")
|
| 166 |
|
| 167 |
status_output = gr.Textbox(label="Run Status / Submission Result", lines=5, interactive=False)
|
| 168 |
# Removed max_rows=10 from DataFrame constructor
|