Update

RSTZZZ · RSTZZZ · commit 10d8f6a3cbff · 2024-12-10T07:49:50.000Z
diff --git a/_posts/papers/2023-07-03-2307.01026.md b/_posts/papers/2023-07-03-2307.01026.md
@@ -8,7 +8,9 @@ tags:
 - Neural Information Processing Systems
 link: https://arxiv.org/abs/2307.01026
 author: Shenyang Huang
-categories: Publications
+categories: 
+- Publications
+- TGB
 
 ---
 
diff --git a/_posts/papers/2024-06-14-2406.09639.md b/_posts/papers/2024-06-14-2406.09639.md
@@ -9,7 +9,9 @@ tags:
 - arXiv.org
 link: https://arxiv.org/abs/2406.09639
 author: Shenyang Huang
-categories: Publications
+categories: 
+- Publications
+- TGB
 
 ---
 
diff --git a/_posts/papers/2024-12-05-2412.04626.md b/_posts/papers/2024-12-05-2412.04626.md
@@ -0,0 +1,29 @@
+---
+title: 'BigDocs: An Open and Permissively-Licensed Dataset for Training Multimodal
+  Models on Document and Code Tasks'
+venue: ''
+names: Juan Rodriguez, Xiangru Jian, Siba Smarak Panigrahi, Tianyu Zhang, Aarash Feizi,
+  Abhay Puri, Akshay Kalkunte, Franccois Savard, Ahmed Masry, Shravan Nayak, Rabiul
+  Awal, Mahsa Massoud, Amirhossein Abaskohi, Zichao Li, Suyuchen Wang, Pierre-Andre
+  Noel, M. L. Richter, Saverio Vadacchino, Shubbam Agarwal, Sanket Biswas, Sara Shanian,
+  Ying Zhang, Noah Bolger, Kurt MacDonald, Simon Fauvel, Sathwik Tejaswi, Srinivas
+  Sunkara, João Monteiro, K. Dvijotham, Torsten Scholak, Nicolas Chapados, Sepideh
+  Kharagani, Sean Hughes, M. Ozsu, Siva Reddy, M. Pedersoli, Y. Bengio, Christopher
+  Pal, I. Laradji, Spandanna Gella, Perouz Taslakian, David Vázquez, Sai Rajeswar
+tags:
+- ''
+link: https://arxiv.org/abs/2412.04626
+author: Aarash Feizi
+categories: Publications
+
+---
+
+*{{ page.names }}*
+
+**{{ page.venue }}**
+
+{% include display-publication-links.html pub=page %}
+
+## Abstract
+
+Multimodal AI has the potential to significantly enhance document-understanding tasks, such as processing receipts, understanding workflows, extracting data from documents, and summarizing reports. Code generation tasks that require long-structured outputs can also be enhanced by multimodality. Despite this, their use in commercial applications is often limited due to limited access to training data and restrictive licensing, which hinders open access. To address these limitations, we introduce BigDocs-7.5M, a high-quality, open-access dataset comprising 7.5 million multimodal documents across 30 tasks. We use an efficient data curation process to ensure our data is high-quality and license-permissive. Our process emphasizes accountability, responsibility, and transparency through filtering rules, traceable metadata, and careful content analysis. Additionally, we introduce BigDocs-Bench, a benchmark suite with 10 novel tasks where we create datasets that reflect real-world use cases involving reasoning over Graphical User Interfaces (GUI) and code generation from images. Our experiments show that training with BigDocs-Bench improves average performance up to 25.8% over closed-source GPT-4o in document reasoning and structured output tasks such as Screenshot2HTML or Image2Latex generation. Finally, human evaluations showed a preference for outputs from models trained on BigDocs over GPT-4o. This suggests that BigDocs can help both academics and the open-source community utilize and improve AI tools to enhance multimodal capabilities and document reasoning. The project is hosted at https://bigdocs.github.io .
diff --git a/_research_directions/temporal-graph-learning.md b/_research_directions/temporal-graph-learning.md
@@ -8,7 +8,7 @@ header:
     overlay_image: /assets/images/research_directions/temporal-graph-learning/banner.webp
 logo_image_path: /assets/images/home/TGL_logo-light.png
 logo_dark_image_path: /assets/images/home/TGL_logo-dark.png
-one-liner: "Advancing the frontier of machine learning on time-evolving graphs to better model and predict dynamic real-world networks and relationships."
+one-liner: "When networks breathe, pulse, and transform, what fundamental truths about complexity and connection can machine learning unveil?"
 excerpt: "Our team brings together experts from fields like AI, data mining, and public health to improve how we analyze networks that change over time. We’re focused on creating better ways to predict outcomes and make decisions, especially in areas like fraud detection and disease tracking."
 
 ---
diff --git a/_research_directions/temporal-graph-learning/tga.md b/_research_directions/temporal-graph-learning/tga.md
@@ -4,8 +4,9 @@ layout: splash
 category: temporal-graph-learning
 order: 2
 header:
-    overlay_filter: linear-gradient(rgba(255, 255, 255, 0.1), rgba(0, 0, 0, 0.5))
+    overlay_filter: 
     overlay_image: /assets/images/research_directions/temporal-graph-learning/TGA.png
+    overlay_size: "contain"
 excerpt: ""
 
 ---
diff --git a/_research_directions/temporal-graph-learning/tgb.md b/_research_directions/temporal-graph-learning/tgb.md
@@ -9,7 +9,8 @@ header:
     overlay_size: "contain"
 excerpt: ""
 
-redirect_to: https://tgb.complexdatalab.com/
+
 ---
 
+TGB is a collection of challenging and diverse benchmark datasets for realistic, reproducible, and robust evaluation for machine learning on temporal graphs. TGB includes both dynamic link and node property prediction tasks and an automated pipeline from dataset downloading, dataloading, evaluation and submission to the TGB leaderboard. Learn more [here](https://tgb.complexdatalab.com/)
 
diff --git a/records/semantic_paper_ids_ignored.json b/records/semantic_paper_ids_ignored.json
@@ -9,6 +9,7 @@
   "056a2ab1b2bb30bc8099660dfe5131e9660d1b90",
   "05d16d7f6dc2fbb571ae3ff7d8145f01c17443e1",
   "05d76a00dd42fd5e75a7222d4479d2a35608c7a3",
+  "05fac44a97728b033f2bbf32f273a8b20b5dce40",
   "06e81ae4a28d9c4c6e79e1ddff0c485a0a178ff1",
   "09a34aad92e6f416c81f47e60de0809616b49cce",
   "0b483b550b21ec42d693fc04a372dbb10dd07019",
