UI: switch to English-only labels/messages; minor cleanup. No logic changes.

app.py

@@ -163,7 +163,7 @@ def _schedule_round_robin_by_test_id(pairs: List[Dict[str, str]], seed: int | No
 
 def load_task(task_name: str):
     if not task_name:
-        raise gr.Error("请先选择任务。")
+        raise gr.Error("Please select a task first.")
 
     rows = _load_task_rows(task_name)
     pairs = _build_image_pairs(rows, task_name)
@@ -183,7 +183,7 @@ def load_task(task_name: str):
         p["swap"] = bool(idx % 2)  # True -> A=B's image; False -> A=A's image
 
     if not pairs:
-        raise gr.Error("没有找到可评测的图片对，请检查数据文件。")
+        raise gr.Error("No valid image pairs found for evaluation. Please check the data.")
 
     return pairs
 
@@ -254,13 +254,13 @@ def _upload_eval_record_to_dataset(task_name: str, row: Dict[str, object]) -> tu
     Returns (ok, message) for UI feedback and debugging.
     """
     if HfApi is None:
-        return False, "huggingface_hub 未安装"
+        return False, "huggingface_hub not installed"
     token = os.environ.get("HF_TOKEN") or os.environ.get("HUGGINGFACEHUB_API_TOKEN")
     repo_id = os.environ.get("EVAL_REPO_ID", "peiranli0930/VisEval")
     if not token:
-        return False, "未找到写入 Token (HF_TOKEN/HUGGINGFACEHUB_API_TOKEN)"
+        return False, "Missing write token (HF_TOKEN/HUGGINGFACEHUB_API_TOKEN)"
     if not repo_id:
-        return False, "未设置 EVAL_REPO_ID"
+        return False, "EVAL_REPO_ID is not set"
     try:
         from huggingface_hub import CommitOperationAdd
 
@@ -277,14 +277,14 @@ def _upload_eval_record_to_dataset(task_name: str, row: Dict[str, object]) -> tu
             operations=operations,
             commit_message=f"Add eval {folder} {row.get('test_id')} {uid}",
         )
-        return True, f"上传成功: {repo_id}/{path_in_repo}"
+        return True, f"Uploaded: {repo_id}/{path_in_repo}"
     except Exception as e:
         # Print to logs for debugging in Space
         try:
             print("[VisArena] Upload to dataset failed:", repr(e))
         except Exception:
             pass
-        return False, f"异常: {type(e).__name__}: {e}"
+        return False, f"Exception: {type(e).__name__}: {e}"
 
 
 def on_task_change(task_name: str, _state_pairs: List[Dict[str, str]]):
@@ -304,7 +304,7 @@ def on_task_change(task_name: str, _state_pairs: List[Dict[str, str]]):
         _resolve_image_path(a_path),
         _resolve_image_path(b_path),
         *default_scores,
-        gr.update(value=f"共 {len(pairs)} 个待评测的图片对。"),
+        gr.update(value=f"Total {len(pairs)} pairs pending evaluation."),
     )
 
 
@@ -345,7 +345,7 @@ def on_submit(
         raise gr.Error("请先选择任务。")
 
     if not pairs:
-        raise gr.Error("当前任务没有加载任何图片对。")
+        raise gr.Error("No image pairs loaded for the current task.")
 
     pair = pairs[index]
     score_map = {
@@ -378,9 +378,9 @@ def on_submit(
     ok_hub, hub_msg = _upload_eval_record_to_dataset(task_name, row)
 
     next_index = min(index + 1, len(pairs) - 1)
-    info = f"已保存 Test ID {pair['test_id']} 的评价结果。"
-    info += " 本地持久化" + ("成功" if ok_local else "失败") + "。"
-    info += " 上传Hub" + ("成功" if ok_hub else "失败") + (f"（{hub_msg}）" if hub_msg else "") + "。"
+    info = f"Saved evaluation for Test ID {pair['test_id']}."
+    info += " Local persistence " + ("succeeded" if ok_local else "failed") + "."
+    info += " Dataset upload " + ("succeeded" if ok_hub else "failed") + (f" ({hub_msg})" if hub_msg else "") + "."
 
     if next_index != index:
         pair = pairs[next_index]
@@ -395,7 +395,7 @@ def on_submit(
             _resolve_image_path(b_path),
             3, 3, 3, 3,
             3, 3, 3, 3,
-            gr.update(value=info + f" 自动跳转到下一组（{next_index + 1}/{len(pairs)}）。"),
+            gr.update(value=info + f" Moved to next pair ({next_index + 1}/{len(pairs)})."),
         )
 
     return (
@@ -406,7 +406,7 @@ def on_submit(
         gr.update(),
         3, 3, 3, 3,
         3, 3, 3, 3,
-        gr.update(value=info + " 已经是最后一组。"),
+        gr.update(value=info + " This is the last pair."),
     )
 
 
@@ -414,7 +414,7 @@ with gr.Blocks(title="VisArena Human Evaluation") as demo:
     gr.Markdown(
         """
         # VisArena Human Evaluation
-        请选择任务并对模型生成的图像进行评分。每项评分范围为 **1（效果极差）** 到 **5（效果极佳）**。
+        Please select a task and rate the generated images. Each score ranges from 1 (poor) to 5 (excellent).
         """
     )
 
@@ -442,21 +442,21 @@ with gr.Blocks(title="VisArena Human Evaluation") as demo:
     # Layout: Original on top, two outputs below with their own sliders
     with gr.Row():
         with gr.Column(scale=12):
-            orig_image = gr.Image(type="filepath", label="原图 Original", interactive=False)
+            orig_image = gr.Image(type="filepath", label="Original", interactive=False)
 
     with gr.Row():
         with gr.Column(scale=6):
-            model1_image = gr.Image(type="filepath", label="模型 A 输出", interactive=False)
-            a_physical_input = gr.Slider(1, 5, value=3, step=1, label="A: 物理交互保真度")
-            a_optical_input = gr.Slider(1, 5, value=3, step=1, label="A: 光学效应准确度")
-            a_semantic_input = gr.Slider(1, 5, value=3, step=1, label="A: 语义/功能对齐度")
-            a_overall_input = gr.Slider(1, 5, value=3, step=1, label="A: 整体真实感")
+            model1_image = gr.Image(type="filepath", label="Output A", interactive=False)
+            a_physical_input = gr.Slider(1, 5, value=3, step=1, label="A: Physical Interaction Fidelity")
+            a_optical_input = gr.Slider(1, 5, value=3, step=1, label="A: Optical Effect Accuracy")
+            a_semantic_input = gr.Slider(1, 5, value=3, step=1, label="A: Semantic/Functional Alignment")
+            a_overall_input = gr.Slider(1, 5, value=3, step=1, label="A: Overall Photorealism")
         with gr.Column(scale=6):
-            model2_image = gr.Image(type="filepath", label="模型 B 输出", interactive=False)
-            b_physical_input = gr.Slider(1, 5, value=3, step=1, label="B: 物理交互保真度")
-            b_optical_input = gr.Slider(1, 5, value=3, step=1, label="B: 光学效应准确度")
-            b_semantic_input = gr.Slider(1, 5, value=3, step=1, label="B: 语义/功能对齐度")
-            b_overall_input = gr.Slider(1, 5, value=3, step=1, label="B: 整体真实感")
+            model2_image = gr.Image(type="filepath", label="Output B", interactive=False)
+            b_physical_input = gr.Slider(1, 5, value=3, step=1, label="B: Physical Interaction Fidelity")
+            b_optical_input = gr.Slider(1, 5, value=3, step=1, label="B: Optical Effect Accuracy")
+            b_semantic_input = gr.Slider(1, 5, value=3, step=1, label="B: Semantic/Functional Alignment")
+            b_overall_input = gr.Slider(1, 5, value=3, step=1, label="B: Overall Photorealism")
 
     submit_button = gr.Button("Submit Evaluation", variant="primary")
     feedback_box = gr.Markdown("")