Add links to paper and code

README.md

@@ -1,51 +1,53 @@
 ---
-pretty_name: BioSum-CUH
-license: other
 language:
-  - en
+- en
+license: other
+size_categories:
+- 10K<n<100K
 task_categories:
-  - summarization
-  - question-answering
-  - token-classification
+- summarization
+- question-answering
+- token-classification
+pretty_name: BioSum-CUH
 tags:
-  - long-context
-  - biography
-  - factuality
-  - hallucination-detection
-  - uncertainty-estimation
-  - token-level-annotations
-  - model-activations
-  - datasets
-size_categories:
-  - 10K<n<100K
+- long-context
+- biography
+- factuality
+- hallucination-detection
+- uncertainty-estimation
+- token-level-annotations
+- model-activations
+- datasets
 configs:
-  - config_name: benchmark
-    default: true
-    data_files:
-      - split: train
-        path: benchmark/train-*.parquet
-      - split: validation
-        path: benchmark/validation-*.parquet
-      - split: test
-        path: benchmark/test-*.parquet
-  - config_name: activations
-    data_files:
-      - split: train
-        path: activations/train-*.parquet
-      - split: detector_eval
-        path: activations/detector_eval-*.parquet
-  - config_name: activation-metadata
-    data_files:
-      - split: train
-        path: activation_metadata/train-*.parquet
-      - split: detector_eval
-        path: activation_metadata/detector_eval-*.parquet
+- config_name: benchmark
+  default: true
+  data_files:
+  - split: train
+    path: benchmark/train-*.parquet
+  - split: validation
+    path: benchmark/validation-*.parquet
+  - split: test
+    path: benchmark/test-*.parquet
+- config_name: activations
+  data_files:
+  - split: train
+    path: activations/train-*.parquet
+  - split: detector_eval
+    path: activations/detector_eval-*.parquet
+- config_name: activation-metadata
+  data_files:
+  - split: train
+    path: activation_metadata/train-*.parquet
+  - split: detector_eval
+    path: activation_metadata/detector_eval-*.parquet
 ---
 
 # BioSum-CUH
 
 **A Biography Summarization Benchmark with Token-Level Correctness, Uncertainty, and Hallucination Annotations**
 
+> **Paper:** [UT-ACA: Uncertainty-Triggered Adaptive Context Allocation for Long-Context Inference](https://huggingface.co/papers/2603.18446)  |  **Code:** [github.com/Tommy307/UT-ACA](https://github.com/Tommy307/UT-ACA)
+
 BioSum-CUH is a benchmark for studying factual generation over biography
 contexts. It combines biography-based question answering and structured
 summarization with token-aligned model predictions, final-layer attention
@@ -506,4 +508,4 @@ citation alongside the repository citation above.
 
 For questions, corrections, or removal requests, open a discussion at:
 
-<https://huggingface.co/datasets/Saria307/biosum-cuh/discussions>
+<https://huggingface.co/datasets/Saria307/biosum-cuh/discussions>