Card: A3 is Apache-2.0, judges also serve training, index files usable without the archives; refresh metadata archive

CHECKSUMS_files.sha256

@@ -1,7 +1,7 @@
 9ba9550ad48438d0836ddab3da480b3b69ffa0aac7b7878b5a0039e7ab429411  DATA_LICENSE
 2c3fd71db4528a5f9b851438de5fcf92e45849c5757fd1f2f5f57a5394d4151e  FORMAT.md
 5e101e63f9df12d911324cd0bd81b4234d959754fa63fad5222a2f18932a07c7  LICENSE
-63cb9b9f4d083823c5ede6de49e7fa5b70809997637bc50c26ebd763622fe444  README.md
+8ad75729390cde8d38d0b59b4f52bf63b8a4a304e3c141a6a3dc79a142df9974  README.md
 59854984853104df5c353e2f681a15fc7924742f9a2e468c29af248dce45ce03  VERSION
 f7904c83447d06cdbc8e467b10d258a0bdf2a1a09a799048611eb752e84d5a66  annotations.csv
 cb1bd5f49ce3cc7e11620c89d5b2516672e021dbb3d29d8eb2af2b75f3471b25  tasks.csv

MobileJudgeBench-v1.0.0-metadata.tar.gz

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:245947a808d698ef92f2f58a5f34167ae6804fd955ee7745b4a82eee8c09ccec
-size 46928
+oid sha256:b17e08c9fc8531afc20454472624471a969f36cf321ba6309f999c668f203c19
+size 46998

README.md

@@ -31,7 +31,7 @@ Paper: [Benchmarking LLM Judges for Mobile Agent Evaluation](https://huggingface
 [github.com/WZq975/MobileJudgeBench](https://github.com/WZq975/MobileJudgeBench)
 
 MobileJudgeBench is the benchmark introduced in [*Benchmarking LLM Judges for Mobile Agent
-Evaluation*](https://huggingface.co/papers/2608.11434). Mobile-agent benchmarks increasingly rely on LLM judges to decide whether an agent
+Evaluation*](https://huggingface.co/papers/2608.11434). Mobile-agent benchmarks and training increasingly rely on LLM judges to decide whether an agent
 completed a task; this dataset lets you measure how reliable such judges are. Every trajectory is
 a full record of one agent attempting one task on an Android emulator (screenshots, accessibility
 trees, actions, the agent's reasoning) together with a human success/failure label.
@@ -51,8 +51,10 @@ The trajectory data is 4.6 GiB (PNG screenshots), distributed as 24 tar archives
 benchmark x agent, each under 1 GiB) listed in `manifest.json` with SHA-256 sums; they are hosted
 on the Hugging Face Hub at
 [huggingface.co/datasets/WesleyWzq/MobileJudgeBench](https://huggingface.co/datasets/WesleyWzq/MobileJudgeBench).
-The index files (`annotations.csv`, `tasks.csv`) are small and sit next to the archives, so
-label-only analyses need no download. The code repository
+The index files (`annotations.csv`, `tasks.csv`) sit next to the archives and are shown in the
+dataset viewer; work that only needs the human labels and task metadata can use these two files
+without downloading the archives (`load_dataset("WesleyWzq/MobileJudgeBench", "annotations")` or
+`"tasks"` with the `datasets` library). The code repository
 ([github.com/WZq975/MobileJudgeBench](https://github.com/WZq975/MobileJudgeBench)) holds the
 downloader, the loader and the paper's Baseline judge:
 
@@ -136,7 +138,7 @@ The tasks and app configurations originate from the source benchmarks, which we
 acknowledge: [AndroidWorld](https://github.com/google-research/android_world) (Apache-2.0),
 [B-MoCA](https://github.com/gimme1dollar/b-moca) (Apache-2.0),
 [SPA-Bench](https://github.com/ai-agents-2030/SPA-Bench) (MIT),
-[A3](https://github.com/YuxiangChai/AITK) (MIT), [AndroidLab](https://github.com/THUDM/Android-Lab)
+[A3](https://github.com/YuxiangChai/AITK) (Apache-2.0), [AndroidLab](https://github.com/THUDM/Android-Lab)
 (MIT) and [AndroidArena](https://github.com/AndroidArenaAgent/AndroidArena) (task definitions
 used for research evaluation with attribution). Agents: AndroidWorld M3A/T3A (Apache-2.0) and
 [UI-TARS](https://github.com/bytedance/UI-TARS) (Apache-2.0).

SHA256SUMS

@@ -22,4 +22,4 @@ df9c211c81d3acee6b9b2988d0a32c33d4316755e76a411ba955c8a5f3caae80  MobileJudgeBen
 6b60e9f5884359f82e2eb1da255a9974acc41e3c538e7b85752f5510ef6ffbe1  MobileJudgeBench-v1.0.0-android_lab-qwen2.5-vl-72b.tar
 e26d8d9edbede169f035c830bf2fbb42b14a8e8e88626b4e0535c2063d38ba5b  MobileJudgeBench-v1.0.0-android_lab-llama-3.3-70b.tar
 8227e117dc847d6f84da54a407a791b9e2b8e5a996e004542af7424ca5c49870  MobileJudgeBench-v1.0.0-android_lab-ui-tars-72b.tar
-245947a808d698ef92f2f58a5f34167ae6804fd955ee7745b4a82eee8c09ccec  MobileJudgeBench-v1.0.0-metadata.tar.gz
+b17e08c9fc8531afc20454472624471a969f36cf321ba6309f999c668f203c19  MobileJudgeBench-v1.0.0-metadata.tar.gz

manifest.json

@@ -1134,8 +1134,8 @@
     "VERSION",
     "FORMAT.md"
    ],
-   "sha256": "245947a808d698ef92f2f58a5f34167ae6804fd955ee7745b4a82eee8c09ccec",
-   "bytes": 46928
+   "sha256": "b17e08c9fc8531afc20454472624471a969f36cf321ba6309f999c668f203c19",
+   "bytes": 46998
   }
  }
 }