Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -26,7 +26,6 @@ class SyntheticDataGenerator:
|
|
26 |
self.original_data = None
|
27 |
|
28 |
def initialize_mostly_ai(self) -> Tuple[bool, str]:
|
29 |
-
"""Initialize Mostly AI SDK"""
|
30 |
if not MOSTLY_AI_AVAILABLE:
|
31 |
return False, "Mostly AI SDK not installed. Please install with: pip install mostlyai[local]"
|
32 |
try:
|
@@ -44,7 +43,6 @@ class SyntheticDataGenerator:
|
|
44 |
batch_size: int = 32,
|
45 |
value_protection: bool = True,
|
46 |
) -> Tuple[bool, str]:
|
47 |
-
"""Train the synthetic data generator"""
|
48 |
if not self.mostly:
|
49 |
return False, "Mostly AI SDK not initialized. Please initialize the SDK first."
|
50 |
try:
|
@@ -63,14 +61,12 @@ class SyntheticDataGenerator:
|
|
63 |
}
|
64 |
]
|
65 |
}
|
66 |
-
|
67 |
self.generator = self.mostly.train(config=train_config)
|
68 |
return True, f"Training completed successfully. Model name: {name}"
|
69 |
except Exception as e:
|
70 |
return False, f"Training failed with error: {str(e)}"
|
71 |
|
72 |
def generate_synthetic_data(self, size: int) -> Tuple[Optional[pd.DataFrame], str]:
|
73 |
-
"""Generate synthetic data"""
|
74 |
if not self.generator:
|
75 |
return None, "No trained generator available. Please train a model first."
|
76 |
try:
|
@@ -82,27 +78,28 @@ class SyntheticDataGenerator:
|
|
82 |
|
83 |
def get_quality_report_file(self) -> Optional[str]:
|
84 |
"""
|
85 |
-
|
86 |
-
|
87 |
"""
|
88 |
if not self.generator:
|
89 |
return None
|
90 |
try:
|
91 |
rep = self.generator.reports(display=False)
|
92 |
|
93 |
-
#
|
94 |
if isinstance(rep, str) and rep.endswith(".zip") and os.path.exists(rep):
|
95 |
return rep
|
96 |
|
97 |
-
#
|
98 |
for attr in ("archive_path", "zip_path", "path", "file_path"):
|
99 |
if hasattr(rep, attr):
|
100 |
p = getattr(rep, attr)
|
101 |
if isinstance(p, str) and os.path.exists(p):
|
102 |
return p
|
103 |
|
104 |
-
#
|
105 |
-
|
|
|
106 |
if hasattr(rep, "save"):
|
107 |
try:
|
108 |
rep.save(target_zip)
|
@@ -118,8 +115,8 @@ class SyntheticDataGenerator:
|
|
118 |
except Exception:
|
119 |
pass
|
120 |
|
121 |
-
#
|
122 |
-
target_txt = "/
|
123 |
with open(target_txt, "w", encoding="utf-8") as f:
|
124 |
f.write(str(rep))
|
125 |
return target_txt
|
@@ -128,21 +125,12 @@ class SyntheticDataGenerator:
|
|
128 |
return None
|
129 |
|
130 |
def estimate_memory_usage(self, df: pd.DataFrame) -> str:
|
131 |
-
"""Estimate memory usage for the dataset"""
|
132 |
if df is None or df.empty:
|
133 |
return "No data available to analyze."
|
134 |
-
|
135 |
memory_mb = df.memory_usage(deep=True).sum() / (1024 * 1024)
|
136 |
rows, cols = len(df), len(df.columns)
|
137 |
estimated_training_mb = memory_mb * 4
|
138 |
-
|
139 |
-
if memory_mb < 100:
|
140 |
-
status = "Good"
|
141 |
-
elif memory_mb < 500:
|
142 |
-
status = "Large"
|
143 |
-
else:
|
144 |
-
status = "Very Large"
|
145 |
-
|
146 |
return f"""
|
147 |
Memory Usage Estimate:
|
148 |
- Data size: {memory_mb:.1f} MB
|
@@ -152,10 +140,12 @@ Memory Usage Estimate:
|
|
152 |
""".strip()
|
153 |
|
154 |
|
155 |
-
#
|
156 |
generator = SyntheticDataGenerator()
|
|
|
|
|
157 |
|
158 |
-
# ----
|
159 |
def initialize_sdk() -> str:
|
160 |
ok, msg = generator.initialize_mostly_ai()
|
161 |
return ("Success: " if ok else "Error: ") + msg
|
@@ -178,62 +168,53 @@ def train_model(
|
|
178 |
|
179 |
|
180 |
def generate_data(size: int) -> Tuple[Optional[pd.DataFrame], str]:
|
181 |
-
|
182 |
-
|
183 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
184 |
|
185 |
|
186 |
-
def create_comparison_plot(original_df: pd.DataFrame, synthetic_df: pd.DataFrame)
|
187 |
if original_df is None or synthetic_df is None:
|
188 |
return None
|
189 |
-
|
190 |
numeric_cols = original_df.select_dtypes(include=[np.number]).columns.tolist()
|
191 |
if not numeric_cols:
|
192 |
return None
|
193 |
-
|
194 |
n_cols = min(3, len(numeric_cols))
|
195 |
n_rows = (len(numeric_cols) + n_cols - 1) // n_cols
|
196 |
-
|
197 |
fig = make_subplots(rows=n_rows, cols=n_cols, subplot_titles=numeric_cols[: n_rows * n_cols])
|
198 |
-
|
199 |
for i, col in enumerate(numeric_cols[: n_rows * n_cols]):
|
200 |
row = i // n_cols + 1
|
201 |
col_idx = i % n_cols + 1
|
202 |
-
|
203 |
-
fig.add_trace(
|
204 |
-
go.Histogram(x=original_df[col], name=f"Original {col}", opacity=0.7, nbinsx=20),
|
205 |
-
row=row,
|
206 |
-
col=col_idx,
|
207 |
-
)
|
208 |
-
fig.add_trace(
|
209 |
-
go.Histogram(x=synthetic_df[col], name=f"Synthetic {col}", opacity=0.7, nbinsx=20),
|
210 |
-
row=row,
|
211 |
-
col=col_idx,
|
212 |
-
)
|
213 |
-
|
214 |
fig.update_layout(title="Original vs Synthetic Data Comparison", height=300 * n_rows, showlegend=True)
|
215 |
return fig
|
216 |
|
217 |
|
218 |
-
def download_csv(df: pd.DataFrame) -> Optional[str]:
|
219 |
-
if df is None or df.empty:
|
220 |
-
return None
|
221 |
-
path = "/mnt/data/synthetic_data.csv"
|
222 |
-
df.to_csv(path, index=False)
|
223 |
-
return path
|
224 |
-
|
225 |
-
|
226 |
# ---- UI ----
|
227 |
def create_interface():
|
228 |
with gr.Blocks(title="MOSTLY AI Synthetic Data Generator", theme=gr.themes.Soft()) as demo:
|
229 |
-
# Header image
|
230 |
gr.Image(
|
231 |
value="https://img.mailinblue.com/8225865/images/content_library/original/6880d164e4e4ea1a183ad4c0.png",
|
232 |
show_label=False,
|
233 |
elem_id="header-image",
|
234 |
)
|
235 |
|
236 |
-
# README
|
237 |
gr.Markdown(
|
238 |
"""
|
239 |
# Synthetic Data SDK by MOSTLY AI Demo Space
|
@@ -289,6 +270,7 @@ def create_interface():
|
|
289 |
train_status = gr.Textbox(label="Training Status", interactive=False)
|
290 |
|
291 |
with gr.Row():
|
|
|
292 |
get_report_btn = gr.DownloadButton("Get Quality Report", variant="secondary")
|
293 |
|
294 |
with gr.Tab("Generate Data"):
|
@@ -302,10 +284,11 @@ def create_interface():
|
|
302 |
|
303 |
synthetic_data = gr.Dataframe(label="Synthetic Data", interactive=False)
|
304 |
with gr.Row():
|
|
|
305 |
download_btn = gr.DownloadButton("Download CSV", variant="secondary")
|
306 |
comparison_plot = gr.Plot(label="Data Comparison")
|
307 |
|
308 |
-
# ----
|
309 |
init_btn.click(initialize_sdk, outputs=[init_status])
|
310 |
|
311 |
train_btn.click(
|
@@ -314,21 +297,18 @@ def create_interface():
|
|
314 |
outputs=[train_status],
|
315 |
)
|
316 |
|
317 |
-
#
|
318 |
-
get_report_btn.click(generator.get_quality_report_file, outputs=
|
319 |
|
320 |
-
# Generate data
|
321 |
generate_btn.click(generate_data, inputs=[gen_size], outputs=[synthetic_data, gen_status])
|
322 |
|
323 |
-
# Update CSV DownloadButton whenever synthetic data changes
|
324 |
-
synthetic_data.change(download_csv, inputs=[synthetic_data], outputs=[download_btn])
|
325 |
-
|
326 |
# Build comparison plot when both datasets are available
|
327 |
-
synthetic_data.change(
|
328 |
-
|
329 |
-
)
|
|
|
330 |
|
331 |
-
#
|
332 |
def process_uploaded_file(file):
|
333 |
if file is None:
|
334 |
return None, "No file uploaded.", gr.update(visible=False)
|
|
|
26 |
self.original_data = None
|
27 |
|
28 |
def initialize_mostly_ai(self) -> Tuple[bool, str]:
|
|
|
29 |
if not MOSTLY_AI_AVAILABLE:
|
30 |
return False, "Mostly AI SDK not installed. Please install with: pip install mostlyai[local]"
|
31 |
try:
|
|
|
43 |
batch_size: int = 32,
|
44 |
value_protection: bool = True,
|
45 |
) -> Tuple[bool, str]:
|
|
|
46 |
if not self.mostly:
|
47 |
return False, "Mostly AI SDK not initialized. Please initialize the SDK first."
|
48 |
try:
|
|
|
61 |
}
|
62 |
]
|
63 |
}
|
|
|
64 |
self.generator = self.mostly.train(config=train_config)
|
65 |
return True, f"Training completed successfully. Model name: {name}"
|
66 |
except Exception as e:
|
67 |
return False, f"Training failed with error: {str(e)}"
|
68 |
|
69 |
def generate_synthetic_data(self, size: int) -> Tuple[Optional[pd.DataFrame], str]:
|
|
|
70 |
if not self.generator:
|
71 |
return None, "No trained generator available. Please train a model first."
|
72 |
try:
|
|
|
78 |
|
79 |
def get_quality_report_file(self) -> Optional[str]:
|
80 |
"""
|
81 |
+
Build/export the quality report and return a file path for immediate download.
|
82 |
+
Uses /tmp for Spaces; tries ZIP, falls back to TXT.
|
83 |
"""
|
84 |
if not self.generator:
|
85 |
return None
|
86 |
try:
|
87 |
rep = self.generator.reports(display=False)
|
88 |
|
89 |
+
# If a string path to a .zip is returned
|
90 |
if isinstance(rep, str) and rep.endswith(".zip") and os.path.exists(rep):
|
91 |
return rep
|
92 |
|
93 |
+
# If object exposes a path-like attribute
|
94 |
for attr in ("archive_path", "zip_path", "path", "file_path"):
|
95 |
if hasattr(rep, attr):
|
96 |
p = getattr(rep, attr)
|
97 |
if isinstance(p, str) and os.path.exists(p):
|
98 |
return p
|
99 |
|
100 |
+
# Try saving/exporting
|
101 |
+
os.makedirs("/tmp", exist_ok=True)
|
102 |
+
target_zip = "/tmp/quality_report.zip"
|
103 |
if hasattr(rep, "save"):
|
104 |
try:
|
105 |
rep.save(target_zip)
|
|
|
115 |
except Exception:
|
116 |
pass
|
117 |
|
118 |
+
# Fallback: stringify into TXT
|
119 |
+
target_txt = "/tmp/quality_report.txt"
|
120 |
with open(target_txt, "w", encoding="utf-8") as f:
|
121 |
f.write(str(rep))
|
122 |
return target_txt
|
|
|
125 |
return None
|
126 |
|
127 |
def estimate_memory_usage(self, df: pd.DataFrame) -> str:
|
|
|
128 |
if df is None or df.empty:
|
129 |
return "No data available to analyze."
|
|
|
130 |
memory_mb = df.memory_usage(deep=True).sum() / (1024 * 1024)
|
131 |
rows, cols = len(df), len(df.columns)
|
132 |
estimated_training_mb = memory_mb * 4
|
133 |
+
status = "Good" if memory_mb < 100 else ("Large" if memory_mb < 500 else "Very Large")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
134 |
return f"""
|
135 |
Memory Usage Estimate:
|
136 |
- Data size: {memory_mb:.1f} MB
|
|
|
140 |
""".strip()
|
141 |
|
142 |
|
143 |
+
# App state
|
144 |
generator = SyntheticDataGenerator()
|
145 |
+
_last_synth_df: Optional[pd.DataFrame] = None # store latest synthetic DF for download
|
146 |
+
|
147 |
|
148 |
+
# ---- Gradio wrappers ----
|
149 |
def initialize_sdk() -> str:
|
150 |
ok, msg = generator.initialize_mostly_ai()
|
151 |
return ("Success: " if ok else "Error: ") + msg
|
|
|
168 |
|
169 |
|
170 |
def generate_data(size: int) -> Tuple[Optional[pd.DataFrame], str]:
|
171 |
+
global _last_synth_df
|
172 |
+
synth_df, message = generator.generate_synthetic_data(size)
|
173 |
+
if synth_df is not None:
|
174 |
+
_last_synth_df = synth_df.copy()
|
175 |
+
return synth_df, f"Success: {message}"
|
176 |
+
else:
|
177 |
+
return None, f"Error: {message}"
|
178 |
+
|
179 |
+
|
180 |
+
def download_csv_now() -> Optional[str]:
|
181 |
+
"""Write the most recent synthetic DF to /tmp and return the path for direct download."""
|
182 |
+
global _last_synth_df
|
183 |
+
if _last_synth_df is None or _last_synth_df.empty:
|
184 |
+
return None
|
185 |
+
os.makedirs("/tmp", exist_ok=True)
|
186 |
+
path = "/tmp/synthetic_data.csv"
|
187 |
+
_last_synth_df.to_csv(path, index=False)
|
188 |
+
return path
|
189 |
|
190 |
|
191 |
+
def create_comparison_plot(original_df: pd.DataFrame, synthetic_df: pd.DataFrame):
|
192 |
if original_df is None or synthetic_df is None:
|
193 |
return None
|
|
|
194 |
numeric_cols = original_df.select_dtypes(include=[np.number]).columns.tolist()
|
195 |
if not numeric_cols:
|
196 |
return None
|
|
|
197 |
n_cols = min(3, len(numeric_cols))
|
198 |
n_rows = (len(numeric_cols) + n_cols - 1) // n_cols
|
|
|
199 |
fig = make_subplots(rows=n_rows, cols=n_cols, subplot_titles=numeric_cols[: n_rows * n_cols])
|
|
|
200 |
for i, col in enumerate(numeric_cols[: n_rows * n_cols]):
|
201 |
row = i // n_cols + 1
|
202 |
col_idx = i % n_cols + 1
|
203 |
+
fig.add_trace(go.Histogram(x=original_df[col], name=f"Original {col}", opacity=0.7, nbinsx=20), row=row, col=col_idx)
|
204 |
+
fig.add_trace(go.Histogram(x=synthetic_df[col], name=f"Synthetic {col}", opacity=0.7, nbinsx=20), row=row, col=col_idx)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
205 |
fig.update_layout(title="Original vs Synthetic Data Comparison", height=300 * n_rows, showlegend=True)
|
206 |
return fig
|
207 |
|
208 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
209 |
# ---- UI ----
|
210 |
def create_interface():
|
211 |
with gr.Blocks(title="MOSTLY AI Synthetic Data Generator", theme=gr.themes.Soft()) as demo:
|
|
|
212 |
gr.Image(
|
213 |
value="https://img.mailinblue.com/8225865/images/content_library/original/6880d164e4e4ea1a183ad4c0.png",
|
214 |
show_label=False,
|
215 |
elem_id="header-image",
|
216 |
)
|
217 |
|
|
|
218 |
gr.Markdown(
|
219 |
"""
|
220 |
# Synthetic Data SDK by MOSTLY AI Demo Space
|
|
|
270 |
train_status = gr.Textbox(label="Training Status", interactive=False)
|
271 |
|
272 |
with gr.Row():
|
273 |
+
# This download button calls a function that returns a file path β download starts immediately
|
274 |
get_report_btn = gr.DownloadButton("Get Quality Report", variant="secondary")
|
275 |
|
276 |
with gr.Tab("Generate Data"):
|
|
|
284 |
|
285 |
synthetic_data = gr.Dataframe(label="Synthetic Data", interactive=False)
|
286 |
with gr.Row():
|
287 |
+
# Same pattern: click β function returns the CSV path β immediate download
|
288 |
download_btn = gr.DownloadButton("Download CSV", variant="secondary")
|
289 |
comparison_plot = gr.Plot(label="Data Comparison")
|
290 |
|
291 |
+
# ---- Events ----
|
292 |
init_btn.click(initialize_sdk, outputs=[init_status])
|
293 |
|
294 |
train_btn.click(
|
|
|
297 |
outputs=[train_status],
|
298 |
)
|
299 |
|
300 |
+
# IMPORTANT: For DownloadButton, do NOT specify outputs β the returned path is auto-downloaded.
|
301 |
+
get_report_btn.click(generator.get_quality_report_file, inputs=None, outputs=None)
|
302 |
|
|
|
303 |
generate_btn.click(generate_data, inputs=[gen_size], outputs=[synthetic_data, gen_status])
|
304 |
|
|
|
|
|
|
|
305 |
# Build comparison plot when both datasets are available
|
306 |
+
synthetic_data.change(create_comparison_plot, inputs=[uploaded_data, synthetic_data], outputs=[comparison_plot])
|
307 |
+
|
308 |
+
# CSV download: return a path from the click handler (no outputs)
|
309 |
+
download_btn.click(download_csv_now, inputs=None, outputs=None)
|
310 |
|
311 |
+
# File upload handler
|
312 |
def process_uploaded_file(file):
|
313 |
if file is None:
|
314 |
return None, "No file uploaded.", gr.update(visible=False)
|