Replicating automatic LLM reviews 4/N
@askalphaxiv
-
PTAS for Game Equilibria Could Redefine PPAD Complexity
By
–
For decades, researchers have debated the existence of a PTAS (polynomial-time approximation scheme) for game equilibria. In this new work from @shb20tsinghua
, the authors propose a PTAS, implying PPAD=FP. Could this overturn the belief that PPAD contains intractable problems? -
Allen AI and UW Release Open-Source Multimodal AI Models
By
–
New work from @allen_ai and @UW
: SOTA multimodal AI models that are truly open-source. The authors have open-sourced their full pipeline, including model weights (Molmo), datasets (PixMo), and training code. Excited to have @inkynumbers and team discussing their work here! -
LLMs Generate More Novel Research Ideas Than Humans
By
–
Can LLMs Generate Novel Research Ideas? This paper from @StanfordAILab performs a human study with 100+ NLP researchers and discovers that LLM-generated ideas are more novel! Excited to have the authors @ChengleiSi @Diyi_Yang and @tatsu_hashimoto here to discuss their work!
-
Chain-of-Thought Prompting: Which Tasks Benefit Most?
By
–
To CoT or not to CoT? Which tasks does Chain-of-thought (CoT) prompting benefit the most? The authors are here to answer your questions! @ZayneSprague @fangcong_y10593 @juand_r_nlp @Dongwei__Jiang @ManyaWadhwa1 @prasann_singhal @lucy_xyzhao @xiye_nlp @kmahowald @gregd_nlp
-
alphaXiv Becomes arXiv Labs Project with Chrome Extension
By
–
Hi Alessandro! We're in the process of becoming an arXiv Labs project, upon which arXiv will include linking to alphaXiv! Until then, we have a Chrome extension with automatic linking when reading papers on arXiv. Feel free to check it out here:
-
Formal domains data integration in LLM pretraining
By
–
5) Are there are other formal or rule-based domains where including its data in LLM pretraining can help?
-
Code Impact in Pretraining: Systematic Study and Analysis
By
–
4) The first work that systematically studies the impact of code in pretraining, inspired by Yoav Goldberg and Yao Fu’s assumptions
-
Code Properties Enhancing Natural Language Task Performance
By
–
3) What are the main properties of code that contribute to the improvement in performance on natural language tasks?
-
Generating High-Quality Synthetic Code Data for Pretraining
By
–
2) How can we generate high-quality synthetic code data to assist in pretraining?