update dataset card(task_id)

README.md

@@ -8,7 +8,6 @@ task_categories:
   - question-answering
 task_ids:
   - multiple-choice-qa
-  - multimodal-question-answering
 pretty_name: TRIAD
 tags:
   - multimodal
@@ -45,7 +44,6 @@ size_categories:
 </div>
 
 ---
-
 ## Overview
 
 TRIAD targets a failure mode common in multimodal evaluation: a model may appear to solve a multimodal task while actually relying on only one dominant modality. In TRIAD, each example is constructed so that the full tri-modal input is needed to identify the intended answer.
@@ -67,7 +65,6 @@ TRIAD targets a failure mode common in multimodal evaluation: a model may appear
 </div>
 
 ---
-
 ## Table of Contents
 
 - [Overview](#overview)
@@ -88,7 +85,6 @@ TRIAD targets a failure mode common in multimodal evaluation: a model may appear
 - [Double-blind Review Notice](#double-blind-review-notice)
 
 ---
-
 ## What Makes TRIAD Different?
 
 TRIAD is designed around **tri-modal irreducibility**.
@@ -113,7 +109,6 @@ The benchmark emphasizes audio cues that cannot be fully captured by a transcrip
 - symbolic sound such as alarms, ringtones, and bells.
 
 ---
-
 ## Repository Layout
 
 ```text
@@ -136,7 +131,6 @@ TRIAD/
 > The exact layout may vary across release versions, but every item contains relative paths to its image and audio files.
 
 ---
-
 ## Data Format
 
 Each item is stored as a JSON object.
@@ -195,7 +189,6 @@ Each item is stored as a JSON object.
 > **Schema note.** Downstream users should treat `options` as the canonical option field name.
 
 ---
-
 ## Loading the Dataset
 
 ### Load JSON metadata
@@ -279,7 +272,6 @@ print(dataset["train"][0])
 </div>
 
 ---
-
 ## Ambiguity Taxonomy
 
 Every item receives one leaf-level ambiguity label for each modality.
@@ -318,7 +310,6 @@ Every item receives one leaf-level ambiguity label for each modality.
 | **Symbolic**      |    20 | Non-linguistic auditory symbols such as alarms, bells, ringtones, or notifications. |
 
 ---
-
 ## Recommended Evaluation Protocol
 
 The standard TRIAD evaluation provides the model with:
@@ -369,7 +360,6 @@ Options:
 
 Answer with a single option key.
 ```
-
 ### Metrics
 
 | Metric               | Description                                                  |
@@ -378,7 +368,6 @@ Answer with a single option key.
 | Group-level accuracy | Accuracy aggregated over scenario groups to reduce over-counting of sister examples. |
 
 ---
-
 ## Intended Uses
 
 TRIAD is intended for:
@@ -391,7 +380,6 @@ TRIAD is intended for:
 - analyzing which ambiguity types are hardest for different model families.
 
 ---
-
 ## Out-of-Scope Uses
 
 TRIAD is **not** intended for:
@@ -409,7 +397,6 @@ TRIAD is **not** intended for:
 > TRIAD is diagnostic by design. It should not be treated as a representative sample of real-world multimodal interactions.
 
 ---
-
 ## Data Collection and Annotation
 
 TRIAD was constructed by designing everyday scenarios in which text, image, and audio jointly determine the intended answer.
@@ -426,7 +413,6 @@ Each item is annotated with:
 Items were reviewed to reduce degenerate cases where the answer can be determined from only one or two modalities.
 
 ---
-
 ## Responsible AI Considerations
 
 <details open>
@@ -466,7 +452,6 @@ The dataset contains English, Chinese, and mixed-language examples. Results may
 </details>
 
 ---
-
 ## Licensing
 
 The license metadata is set to `other` because TRIAD may include different license terms for metadata, annotations, images, and audio.
@@ -483,7 +468,6 @@ Users should consult the accompanying attribution metadata before redistribution
 During double-blind review, author and institution information is anonymized. Final licensing and attribution details will be completed in the camera-ready release.
 
 ---
-
 ## Citation
 
 Citation information will be added after the double-blind review period.
@@ -498,7 +482,6 @@ Citation information will be added after the double-blind review period.
 ```
 
 ---
-
 ## Maintenance
 
 The dataset will be versioned. Errata, corrected annotations, and future extensions will be documented in the repository release history.
@@ -506,7 +489,6 @@ The dataset will be versioned. Errata, corrected annotations, and future extensi
 For the anonymous review version, contact information is withheld to preserve double-blind review. A public maintainer contact will be added after the review period.
 
 ---
-
 ## Double-blind Review Notice
 
 This repository is prepared for anonymous peer review. Please do not infer author identity from repository ownership, commit metadata, or temporary hosting information. Any non-anonymous information will be added after the review period.