Add Cornstarch

source/_data/SymbioticLab.bib

@@ -2459,4 +2459,20 @@ @Article{branchandbrowse:acl26
   publist_abstract = {
 Autonomous web agents powered by large language models (LLMs) show strong potential for performing goal-oriented tasks such as information retrieval, report generation, and online transactions. These agents mark a key step toward practical embodied reasoning in open web environments. However, existing approaches remain limited in reasoning depth and efficiency: vanilla linear methods fail at multi-step reasoning and lack effective backtracking, while other search strategies are coarse-grained and computationally costly. We introduce Branch-and-Browse, a fine-grained web agent framework that unifies structured reasoning-acting, contextual memory, and efficient execution. It (i) employs explicit subtask management with tree-structured exploration for controllable multi-branch reasoning, (ii) bootstraps exploration through efficient web state replay with background reasoning, and (iii) leverages a page action memory to share explored actions within and across sessions. On the WebArena benchmark, Branch-and-Browse achieves a task success rate of 35.8% and reduces execution time by up to 40.4% relative to state-of-the-art methods. These results demonstrate that Branch-and-Browse is a reliable and efficient framework for LLM-based web agents.
   }
-} 
+} 
+
+@InProceedings{cornstarch:icml26,
+  author        = {Insu Jang and Runyu Lu and Nikhil Bansal and Ang Chen and Mosharaf Chowdhury},
+  title         = {Efficient Distributed {MLLM} Training with {Cornstarch}},
+  year          = {2026},
+  month         = {July},
+  publist_confkey = {ICML'26},
+  publist_link = {paper || cornstarch-icml26.pdf},
+  publist_link = {code || https://github.com/Cornstarch-org/Cornstarch},
+  publist_topic = {Systems + AI},
+  publist_abstract = {
+Multimodal large language models (MLLMs) extend the capabilities of large language models (LLMs) by combining heterogeneous model architectures to handle diverse modalities like images and audio. However, this inherent heterogeneity in MLLM model structure and data types makes makeshift extensions to existing LLM training frameworks unsuitable for efficient MLLM training, especially in distributed training.
+
+In this paper, we present Cornstarch, an efficient distributed MLLM training framework that contemplates MLLM's unique characteristics in both model and data parallelization. Cornstarch introduces frozen-aware pipeline parallelism and workload-balanced context parallelism to improve MLLM training throughput. Our extensive evaluation shows that Cornstarch outperforms state-of-the-art solutions by 2.26x on average in terms of MLLM training throughput.
+  }
+}

source/publications/index.md

@@ -442,6 +442,11 @@ venues:
   ICML:
     category: Conferences
     occurrences:
+    - key: ICML'26
+      name: Forty-Third International Conference on Machine Learning
+      date: 2026-07-06
+      url: https://icml.cc/Conferences/2026
+      acceptance: 26.56%
     - key: ICML'22
       name: Thirty-ninth International Conference on Machine Learning
       date: 2022-07-17