final

app.py

@@ -311,7 +311,7 @@ INSTRUCTION_MD = """
 For each video, your job is to rate:
 
 1. **Action Consistency** - how well the person’s action matches the action specified as the "**Expected action**".  
-2. **Physical Plausibility** - how natural and physically possible the motion looks.
+2. **Temporal Coherence** - how natural and physically possible the motion looks.
 
 Some things to keep in mind:
 - Try to **focus only** on the expected action and motion quality, and do **not** judge **video quality**, **attractiveness**, **background**, **camera motion**, or **objects**.
@@ -720,7 +720,7 @@ with gr.Blocks(css=CLEAN_BG_CSS) as demo:
     page_intro = gr.Group(visible=True, elem_id="intro")
     with page_intro:
         # gr.Markdown("## 🎯 Action Consistency Human Evaluation")
-        gr.Markdown("## 🎯 Human Evaluation: Action Consistency & Physical Plausibility")
+        gr.Markdown("## 🎯 Human Evaluation: Action Consistency & Temporal Coherence")
 
         gr.Markdown(INSTRUCTION_MD)
 
@@ -868,7 +868,7 @@ with gr.Blocks(css=CLEAN_BG_CSS) as demo:
                 )
                 score_phys = gr.Slider(
                     minimum=0.0, maximum=10.0, step=0.1, value=5.0,
-                    label="Physical Plausibility (0.0 - 10.0)"
+                    label="Temporal Coherence (0.0 - 10.0)"
                 )
 
                 save_next = gr.Button("💾 Save & Next ▶", variant="secondary", interactive=False)