Spaces:
Runtime error
Runtime error
File size: 11,947 Bytes
9511c28 5f74c91 63dc66a b271aab 439407c b271aab 439407c | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 | # # app.py
# """
# Gradio application entrypoint for Hugging Face Spaces.
# """
# import os
# import tempfile
# import pandas as pd
# import gradio as gr
# from evaluator import evaluate_dataframe
# from synthetic_data import generate_synthetic_dataset
# # Helper to save uploaded file to local temp path (gradio File gives a NamedTemporaryFile-like object)
# def save_uploaded(file_obj):
# if not file_obj:
# return None
# # file_obj can be a dictionary or a file-like object depending on Gradio version
# try:
# path = file_obj.name
# return path
# except Exception:
# # fallback: write bytes to temp file
# data = file_obj.read()
# suffix = ".csv" if file_obj.name.endswith(".csv") else ".json"
# fd, tmp = tempfile.mkstemp(suffix=suffix)
# with os.fdopen(fd, "wb") as f:
# f.write(data)
# return tmp
# def load_file_to_df(path):
# if path is None:
# return None
# # Try CSV
# try:
# if path.endswith(".csv"):
# return pd.read_csv(path)
# # JSONL
# try:
# return pd.read_json(path, lines=True)
# except ValueError:
# return pd.read_json(path)
# except Exception as e:
# # As last resort, raise
# raise e
# def run_evaluation(file_obj):
# # If no file provided, use synthetic demo
# if file_obj is None:
# df = generate_synthetic_dataset(num_agents=3, num_samples=12)
# else:
# path = save_uploaded(file_obj)
# df = load_file_to_df(path)
# # Ensure required columns exist; otherwise, attempt to map common alternatives
# if df is None:
# return None, "No data loaded", None
# # Try to normalize column names
# cols = {c.lower(): c for c in df.columns}
# # rename common variants
# rename_map = {}
# for k in ["prompt", "response", "task", "agent", "reference"]:
# if k not in cols:
# # try variants
# if k == "reference":
# for alt in ["answer", "ground_truth", "ref"]:
# if alt in cols:
# rename_map[cols[alt]] = k
# break
# else:
# for alt in [k, k.capitalize(), k.upper()]:
# if alt.lower() in cols:
# rename_map[cols[alt.lower()]] = k
# if rename_map:
# df = df.rename(columns=rename_map)
# metrics_df, images, leaderboard = evaluate_dataframe(df)
# # Prepare gallery (list of image file paths). Gradio Gallery accepts list of image paths or PIL images.
# gallery_items = [p for (p, caption) in images]
# captions = [caption for (p, caption) in images]
# # Save a CSV report for download
# out_csv = "/tmp/eval_results.csv"
# metrics_df.to_csv(out_csv, index=False)
# return (gallery_items, captions), metrics_df, leaderboard
# # Build Gradio UI
# with gr.Blocks() as demo:
# gr.Markdown("# Agentic Evaluation Framework")
# gr.Markdown(
# "Upload a CSV/JSON/JSONL with columns: `prompt,response,task,agent,reference` (reference optional). "
# "If no file is uploaded, a small synthetic demo will run."
# )
# with gr.Row():
# file_input = gr.File(label="Upload CSV / JSON / JSONL (optional)", file_types=[".csv", ".json", ".jsonl"])
# run_btn = gr.Button("Run Evaluation")
# download_report = gr.File(label="Download CSV Report")
# # ✅ Fixed Gallery (removed .style, added columns=2)
# gallery = gr.Gallery(
# label="Visualization Outputs",
# columns=2,
# height="auto"
# )
# table = gr.Dataframe(headers=None, label="Per-example Metrics (detailed)")
# leaderboard = gr.Dataframe(headers=None, label="Leaderboard (Avg Final Score per Agent & Task)")
# def on_run(file_in):
# (gallery_items, captions), metrics_df, lb = run_evaluation(file_in)
# # Save gallery captions mapping into a simple list of tuples for Gradio gallery (path, caption)
# gallery_display = []
# for i, p in enumerate(gallery_items):
# caption = captions[i] if i < len(captions) else ""
# gallery_display.append((p, caption))
# return gallery_display, metrics_df, lb
# run_btn.click(fn=on_run, inputs=[file_input], outputs=[gallery, table, leaderboard])
# gr.Markdown("## Usage tips\n- Columns: `prompt,response,task,agent,reference` (case-insensitive). "
# "- `reference` can be empty but accuracy/hallucination will be weaker.\n"
# "- Visualization images are available in the Gallery and a CSV report is downloadable.")
# demo.launch()
# app.py
# """
# Gradio application entrypoint for Hugging Face Spaces.
# """
# import os
# import tempfile
# import pandas as pd
# import gradio as gr
# from evaluation import evaluate_dataframe # ✅ updated import
# from synthetic_data import generate_synthetic_dataset
# # Helper to save uploaded file
# def save_uploaded(file_obj):
# if not file_obj:
# return None
# try:
# return file_obj.name
# except Exception:
# data = file_obj.read()
# suffix = ".csv" if file_obj.name.endswith(".csv") else ".json"
# fd, tmp = tempfile.mkstemp(suffix=suffix)
# with os.fdopen(fd, "wb") as f:
# f.write(data)
# return tmp
# def load_file_to_df(path):
# if path is None:
# return None
# try:
# if path.endswith(".csv"):
# return pd.read_csv(path)
# try:
# return pd.read_json(path, lines=True)
# except ValueError:
# return pd.read_json(path)
# except Exception as e:
# raise e
# def run_evaluation(file_obj):
# if file_obj is None:
# df = generate_synthetic_dataset(num_agents=3, num_samples=12)
# else:
# path = save_uploaded(file_obj)
# df = load_file_to_df(path)
# if df is None:
# return None, "No data loaded", None
# # Normalize column names
# cols = {c.lower(): c for c in df.columns}
# rename_map = {}
# for k in ["prompt", "response", "task", "agent", "reference"]:
# if k not in cols:
# if k == "reference":
# for alt in ["answer", "ground_truth", "ref"]:
# if alt in cols:
# rename_map[cols[alt]] = k
# break
# else:
# for alt in [k, k.capitalize(), k.upper()]:
# if alt.lower() in cols:
# rename_map[cols[alt.lower()]] = k
# if rename_map:
# df = df.rename(columns=rename_map)
# metrics_df, images, leaderboard = evaluate_dataframe(df)
# gallery_items = [p for (p, caption) in images]
# captions = [caption for (p, caption) in images]
# out_csv = "/tmp/eval_results.csv"
# metrics_df.to_csv(out_csv, index=False)
# return (gallery_items, captions), metrics_df, leaderboard
# # Build Gradio UI
# with gr.Blocks() as demo:
# gr.Markdown("# Agentic Evaluation Framework")
# gr.Markdown(
# "Upload a CSV/JSON/JSONL with columns: `prompt,response,task,agent,reference`. "
# "If no file is uploaded, a synthetic demo will run."
# )
# with gr.Row():
# file_input = gr.File(label="Upload CSV/JSON/JSONL", file_types=[".csv", ".json", ".jsonl"])
# run_btn = gr.Button("Run Evaluation")
# download_report = gr.File(label="Download CSV Report")
# gallery = gr.Gallery(label="Visualization Outputs", columns=2, height="auto")
# table = gr.Dataframe(headers=None, label="Per-example Metrics (detailed)")
# leaderboard = gr.Dataframe(headers=None, label="Leaderboard (Avg Score per Agent & Task)")
# def on_run(file_in):
# (gallery_items, captions), metrics_df, lb = run_evaluation(file_in)
# gallery_display = [(p, captions[i] if i < len(captions) else "") for i, p in enumerate(gallery_items)]
# return gallery_display, metrics_df, lb
# run_btn.click(fn=on_run, inputs=[file_input], outputs=[gallery, table, leaderboard])
# gr.Markdown("## Tips\n- Columns: `prompt,response,task,agent,reference` (case-insensitive). "
# "- `reference` optional.\n- Download CSV report after evaluation.")
# demo.launch()
"""
Gradio application entrypoint for Hugging Face Spaces.
"""
import os
import tempfile
import pandas as pd
import gradio as gr
from evaluator import evaluate_dataframe
from synthetic_data import generate_synthetic_dataset
# -----------------------------
# File Handling
# -----------------------------
def save_uploaded(file_obj):
if not file_obj:
return None
try:
return file_obj.name
except Exception:
data = file_obj.read()
suffix = ".csv" if file_obj.name.endswith(".csv") else ".json"
fd, tmp = tempfile.mkstemp(suffix=suffix)
with os.fdopen(fd, "wb") as f:
f.write(data)
return tmp
def load_file_to_df(path):
if path is None:
return None
try:
if path.endswith(".csv"):
return pd.read_csv(path)
try:
return pd.read_json(path, lines=True)
except ValueError:
return pd.read_json(path)
except Exception as e:
raise e
# -----------------------------
# Evaluation Pipeline
# -----------------------------
def run_evaluation(file_obj):
if file_obj is None:
df = generate_synthetic_dataset(num_agents=3, num_samples=12)
else:
path = save_uploaded(file_obj)
df = load_file_to_df(path)
if df is None:
return None, "No data loaded", None
# Normalize column names
cols = {c.lower(): c for c in df.columns}
rename_map = {}
for k in ["task_id", "prompt", "response", "agent", "reference"]:
if k not in cols:
for alt in [k, k.capitalize(), k.upper()]:
if alt.lower() in cols:
rename_map[cols[alt.lower()]] = k
break
if rename_map:
df = df.rename(columns=rename_map)
metrics_df, images, leaderboard = evaluate_dataframe(df)
gallery_items = [p for (p, caption) in images]
captions = [caption for (p, caption) in images]
# Save CSV report
out_csv = "/tmp/eval_results.csv"
metrics_df.to_csv(out_csv, index=False)
return (gallery_items, captions), metrics_df, leaderboard
# -----------------------------
# Gradio UI
# -----------------------------
with gr.Blocks() as demo:
gr.Markdown("# 🧪 Agentic Evaluation Framework")
gr.Markdown(
"Upload a CSV/JSON/JSONL with columns: "
"`task_id,prompt,response,agent,reference`. "
"If no file is uploaded, a small synthetic demo will run."
)
with gr.Row():
file_input = gr.File(label="Upload CSV/JSON/JSONL", file_types=[".csv", ".json", ".jsonl"])
run_btn = gr.Button("Run Evaluation")
download_report = gr.File(label="Download CSV Report")
gallery = gr.Gallery(label="Visualization Outputs", columns=2, height="auto")
table = gr.Dataframe(headers=None, label="Per-example Metrics (detailed)")
leaderboard = gr.Dataframe(headers=None, label="Leaderboard (Avg Final Score per Agent)")
def on_run(file_in):
(gallery_items, captions), metrics_df, lb = run_evaluation(file_in)
gallery_display = [(p, captions[i] if i < len(captions) else "") for i, p in enumerate(gallery_items)]
return gallery_display, metrics_df, lb
run_btn.click(fn=on_run, inputs=[file_input], outputs=[gallery, table, leaderboard])
gr.Markdown("## Tips\n- Ensure columns: `task_id,prompt,response,agent,reference` "
"(case-insensitive).\n- Visualization images in Gallery.\n- Download CSV after evaluation.")
demo.launch()
|