Add manifest (metadata.json)

metadata.json

@@ -9,7 +9,7 @@
       "modalities": [
         "single_image_start"
       ],
-      "task_type": "vqa",
+      "task_type": "free_form",
       "score_type": "llm_judge",
       "score_params": {
         "rubric": "scale",
@@ -20,10 +20,15 @@
         "official_judge_model": "gpt-4"
       },
       "score_protocol": {
-        "reference": "Official eval script https://huggingface.co/datasets/Shengcao1006/MMHal-Bench/blob/main/eval_gpt4.py — GPT-4 rates each response 0-6 (informativeness x hallucination rubric: 0-2 = hallucination present, 3-6 = none) given image_content object list, question, and standard human answer. Not implemented in VLMEvalKit or lmms-eval.",
-        "note": "Official headline = average rating (0-6) AND hallucination rate = fraction of responses rated 0-2, with per-question-type breakdown (8 types x 12 questions, 96 rows — matches published num_rows_total=96). Per-sample binary grading deviates. mm-eval copy lacks the `image_content` object list the official judge template requires; per-row `question_type` extra carries MMHal categories (attribute/relation/holistic/other...), NOT scorer vocabulary — scorer must not treat it as a question_type signal."
+        "reference": "Official eval script https://huggingface.co/datasets/Shengcao1006/MMHal-Bench/blob/main/eval_gpt4.py (lines 94-96: judge template consumes image_content, question, gt_answer, model_answer) — GPT-4 rates each response 0-6 (informativeness x hallucination rubric: 0-2 = hallucination present, 3-6 = none). Not implemented in VLMEvalKit or lmms-eval.",
+        "note": "Official headline = average rating (0-6) AND hallucination rate = fraction of responses rated 0-2, with per-question-type breakdown (8 types x 12 questions, 96 rows). Per-sample binary grading deviates. Rows carry the image_content object list (extra) that the official judge template interpolates. Per-row question_type extra carries MMHal categories (attribute/relation/holistic/other...), NOT scorer vocabulary — scorer must not treat it as a question_type signal."
       },
       "prompt_template": "<image>{{ question }}",
+      "prompt_template_source": {
+        "origin": "official",
+        "reference": "https://github.com/open-compass/VLMEvalKit/blob/main/vlmeval/dataset/image_vqa.py (MMHal-Bench — judge-scored; bare question)",
+        "notes": "Tier 3: MMHal-Bench judge-scored eval: bare question per the official MMHal-Bench paper."
+      },
       "mapping_from_source": {
         "media": {
           "from": "image",
@@ -32,18 +37,21 @@
           "max_items": 1
         },
         "id": {
-          "from": "id"
+          "from": "image_id"
         },
         "question": {
           "from": "question"
         },
         "answer": {
-          "from": "answer",
+          "from": "gt_answer",
           "optional": true
         },
         "extra": {
           "question_type": {
             "from": "question_type"
+          },
+          "image_content": {
+            "from": "image_content"
           }
         },
         "source": {
@@ -52,11 +60,6 @@
             "test": "https://huggingface.co/datasets/Shengcao1006/MMHal-Bench"
           }
         }
-      },
-      "prompt_template_source": {
-        "origin": "official",
-        "reference": "https://github.com/open-compass/VLMEvalKit/blob/main/vlmeval/dataset/image_vqa.py (MMHal-Bench — judge-scored; bare question)",
-        "notes": "Tier 3: MMHal-Bench judge-scored eval: bare question per the official MMHal-Bench paper."
       }
     }
   }