Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -17,7 +17,7 @@ hf_logging.set_verbosity_error()
|
|
17 |
|
18 |
MODEL_NAME = "bert-base-uncased"
|
19 |
DEVICE = "cpu"
|
20 |
-
SAVE_DIR = "μ μ₯μ μ₯1"
|
21 |
LAYER_ID = 4
|
22 |
SEED = 0
|
23 |
CLF_NAME = "linear"
|
@@ -133,7 +133,6 @@ def analyze_sentence_for_gradio(sentence_text, top_k_value):
|
|
133 |
error_html = f"<p style='color:red;'>Initialization Error: {html.escape(MODEL_LOADING_ERROR_MESSAGE)}</p>"
|
134 |
empty_df = pd.DataFrame(columns=['token', 'score'])
|
135 |
empty_fig = create_empty_plotly_figure("Model Loading Failed")
|
136 |
-
# gr.Labelμ λν μ€λ₯ λ°νκ° μμ (λ¨μ λμ
λ리 λλ λ¬Έμμ΄)
|
137 |
error_label_output = {"Status": "Error", "Message": "Model Loading Failed. Check logs."}
|
138 |
return error_html, [], "Model Loading Failed", error_label_output, [], empty_df, empty_fig
|
139 |
|
@@ -215,8 +214,7 @@ def analyze_sentence_for_gradio(sentence_text, top_k_value):
|
|
215 |
predicted_class_label_str = CLASS_LABEL_MAP.get(pred_idx, f"Unknown Index ({pred_idx})")
|
216 |
|
217 |
prediction_summary_text = f"Predicted Class: {predicted_class_label_str}\nProbability: {pred_prob_val:.3f}"
|
218 |
-
|
219 |
-
prediction_details_for_label = {predicted_class_label_str: float(f"{pred_prob_val:.3f}")} # νλ₯ κ°μ floatμΌλ‘ μ λ¬
|
220 |
|
221 |
pca_fig = create_empty_plotly_figure("PCA Plot N/A\n(Not enough non-special tokens for 3D)")
|
222 |
non_special_token_indices = [idx for idx, token_id in enumerate(input_ids[0,:len(actual_tokens)].tolist())
|
@@ -244,11 +242,10 @@ def analyze_sentence_for_gradio(sentence_text, top_k_value):
|
|
244 |
print(f"analyze_sentence_for_gradio error: {e}\n{tb_str}")
|
245 |
empty_df = pd.DataFrame(columns=['token', 'score'])
|
246 |
empty_fig = create_empty_plotly_figure("Analysis Error")
|
247 |
-
# gr.Labelμ λν μ€λ₯ λ°νκ° μμ
|
248 |
error_label_output = {"Status": "Error", "Message": f"Analysis failed: {str(e)}"}
|
249 |
return error_html, [], "Analysis Failed", error_label_output, [], empty_df, empty_fig
|
250 |
|
251 |
-
# --- Gradio UI Definition (
|
252 |
theme = gr.themes.Monochrome(
|
253 |
primary_hue=gr.themes.colors.blue,
|
254 |
secondary_hue=gr.themes.colors.sky,
|
@@ -265,6 +262,7 @@ with gr.Blocks(title="AI Sentence Analyzer XAI π", theme=theme, css=".gradio-
|
|
265 |
gr.Markdown("Analyze English sentences to understand BERT model predictions through various XAI visualization techniques. "
|
266 |
"Explore token importance and their distribution in the embedding space.")
|
267 |
|
|
|
268 |
with gr.Row(equal_height=False):
|
269 |
with gr.Column(scale=1, min_width=350):
|
270 |
with gr.Group():
|
@@ -276,32 +274,48 @@ with gr.Blocks(title="AI Sentence Analyzer XAI π", theme=theme, css=".gradio-
|
|
276 |
with gr.Column(scale=2):
|
277 |
with gr.Accordion("π― Prediction Outcome", open=True):
|
278 |
output_prediction_summary = gr.Textbox(label="Prediction Summary", lines=2, interactive=False)
|
279 |
-
output_prediction_details = gr.Label(label="Prediction Details & Confidence")
|
280 |
with gr.Accordion("β Top-K Important Tokens (Table)", open=True):
|
281 |
output_top_tokens_df = gr.DataFrame(headers=["Token", "Score"], label="Most Important Tokens",
|
282 |
row_count=(1,"dynamic"), col_count=(2,"fixed"), interactive=False, wrap=True)
|
|
|
283 |
|
284 |
-
|
285 |
-
|
286 |
-
|
287 |
-
|
288 |
-
|
289 |
-
|
290 |
-
|
291 |
-
|
292 |
-
|
293 |
-
|
294 |
-
|
295 |
-
|
296 |
-
|
297 |
-
|
298 |
-
|
299 |
-
|
300 |
-
|
301 |
-
|
302 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
303 |
|
304 |
-
gr.Markdown("---")
|
|
|
305 |
gr.Examples(
|
306 |
examples=[
|
307 |
["This movie is an absolute masterpiece, captivating from start to finish.", 5],
|
|
|
17 |
|
18 |
MODEL_NAME = "bert-base-uncased"
|
19 |
DEVICE = "cpu"
|
20 |
+
SAVE_DIR = "μ μ₯μ μ₯1" # This folder name is from your setup
|
21 |
LAYER_ID = 4
|
22 |
SEED = 0
|
23 |
CLF_NAME = "linear"
|
|
|
133 |
error_html = f"<p style='color:red;'>Initialization Error: {html.escape(MODEL_LOADING_ERROR_MESSAGE)}</p>"
|
134 |
empty_df = pd.DataFrame(columns=['token', 'score'])
|
135 |
empty_fig = create_empty_plotly_figure("Model Loading Failed")
|
|
|
136 |
error_label_output = {"Status": "Error", "Message": "Model Loading Failed. Check logs."}
|
137 |
return error_html, [], "Model Loading Failed", error_label_output, [], empty_df, empty_fig
|
138 |
|
|
|
214 |
predicted_class_label_str = CLASS_LABEL_MAP.get(pred_idx, f"Unknown Index ({pred_idx})")
|
215 |
|
216 |
prediction_summary_text = f"Predicted Class: {predicted_class_label_str}\nProbability: {pred_prob_val:.3f}"
|
217 |
+
prediction_details_for_label = {predicted_class_label_str: float(f"{pred_prob_val:.3f}")}
|
|
|
218 |
|
219 |
pca_fig = create_empty_plotly_figure("PCA Plot N/A\n(Not enough non-special tokens for 3D)")
|
220 |
non_special_token_indices = [idx for idx, token_id in enumerate(input_ids[0,:len(actual_tokens)].tolist())
|
|
|
242 |
print(f"analyze_sentence_for_gradio error: {e}\n{tb_str}")
|
243 |
empty_df = pd.DataFrame(columns=['token', 'score'])
|
244 |
empty_fig = create_empty_plotly_figure("Analysis Error")
|
|
|
245 |
error_label_output = {"Status": "Error", "Message": f"Analysis failed: {str(e)}"}
|
246 |
return error_html, [], "Analysis Failed", error_label_output, [], empty_df, empty_fig
|
247 |
|
248 |
+
# --- Gradio UI Definition (Tabs removed, visualizations shown sequentially or in rows) ---
|
249 |
theme = gr.themes.Monochrome(
|
250 |
primary_hue=gr.themes.colors.blue,
|
251 |
secondary_hue=gr.themes.colors.sky,
|
|
|
262 |
gr.Markdown("Analyze English sentences to understand BERT model predictions through various XAI visualization techniques. "
|
263 |
"Explore token importance and their distribution in the embedding space.")
|
264 |
|
265 |
+
# Inputs and Summary Outputs Row
|
266 |
with gr.Row(equal_height=False):
|
267 |
with gr.Column(scale=1, min_width=350):
|
268 |
with gr.Group():
|
|
|
274 |
with gr.Column(scale=2):
|
275 |
with gr.Accordion("π― Prediction Outcome", open=True):
|
276 |
output_prediction_summary = gr.Textbox(label="Prediction Summary", lines=2, interactive=False)
|
277 |
+
output_prediction_details = gr.Label(label="Prediction Details & Confidence")
|
278 |
with gr.Accordion("β Top-K Important Tokens (Table)", open=True):
|
279 |
output_top_tokens_df = gr.DataFrame(headers=["Token", "Score"], label="Most Important Tokens",
|
280 |
row_count=(1,"dynamic"), col_count=(2,"fixed"), interactive=False, wrap=True)
|
281 |
+
gr.Markdown("---") # Separator
|
282 |
|
283 |
+
# Visualization Section Title
|
284 |
+
gr.Markdown("## π Detailed Visualizations")
|
285 |
+
|
286 |
+
# HTML Highlight (Custom) - Full Width
|
287 |
+
with gr.Group():
|
288 |
+
gr.Markdown("### π¨ HTML Highlight (Custom)")
|
289 |
+
output_html_visualization = gr.HTML(label="Token Importance (Gradient x Input based)")
|
290 |
+
|
291 |
+
# Highlighted Text (Gradio) - Full Width
|
292 |
+
with gr.Group():
|
293 |
+
gr.Markdown("### ποΈ Highlighted Text (Gradio)")
|
294 |
+
output_highlighted_text = gr.HighlightedText(
|
295 |
+
label="Token Importance (Score: 0-1)",
|
296 |
+
show_legend=True,
|
297 |
+
combine_adjacent=False
|
298 |
+
)
|
299 |
+
|
300 |
+
# BarPlot and PCA Plot Side-by-Side
|
301 |
+
with gr.Row():
|
302 |
+
with gr.Column(scale=1, min_width=400): # Adjusted min_width for BarPlot
|
303 |
+
with gr.Group():
|
304 |
+
gr.Markdown("### π Top-K Bar Plot")
|
305 |
+
output_top_tokens_barplot = gr.BarPlot(
|
306 |
+
label="Top-K Token Importance Scores",
|
307 |
+
x="token",
|
308 |
+
y="score",
|
309 |
+
tooltip=['token', 'score'],
|
310 |
+
min_width=300 # BarPlot itself can define min_width
|
311 |
+
)
|
312 |
+
with gr.Column(scale=1, min_width=400): # Adjusted min_width for PCA
|
313 |
+
with gr.Group():
|
314 |
+
gr.Markdown("### π Token Embeddings 3D PCA (Interactive)")
|
315 |
+
output_pca_plot = gr.Plot(label="3D PCA of Token Embeddings (Colored by Importance Score)")
|
316 |
|
317 |
+
gr.Markdown("---") # Separator
|
318 |
+
|
319 |
gr.Examples(
|
320 |
examples=[
|
321 |
["This movie is an absolute masterpiece, captivating from start to finish.", 5],
|