BEYOND WEISFEILER-LEHMAN: A QUANTITATIVE FRAMEWORK FOR GNN EXPRESSIVENESS (ICLR 2024)

Abstract Challenges: WL hierarchy has notatble limitations: coarse, qualitative, and may not well reflect practical requirements They propose a novel framework for quantitatively studying the expressiveness of GNN architectures, addressing the above limitations. Introduction One of the most prominent drawbacks of popular GNNs lie in the limited expressive power. MPNN is bounded by the 1-dimensional …

OPENREVIEWER: MITIGATING CHALLENGES IN LLM REVIEWING (ICLR 2024 Submission)

Abstract Human reviews are slow and of variable quality. Therefore, some people focus on using LLM to do reviewing. Main Challenges: 1. risk of misuse 2. inflated review scores 3. overconfident ratings 4. skewed score distributions 5. limited prompt length Their method: without prompt engineering by using LLM watermarking to mark LLM-generated reviews classifying and …

LLM Agent Survey Reading: The Rise and Potential of Large Language Model Based Agents: A Survey

Abstract: Definition of AI agents: artificial entities that sense their environment, make decisions, and take actions. A general framework that can be tailored for different applications. Three scenarios: single-agent, multi-agent, and human-agent cooperation. Exploring the behavior and personality of LLM-based agents. Background A general conceptual framework for the LLM-based agents with three key parts: brain, …

Generative Agents: Interactive Simulacra of Human Behavior

Abstract: Generative agents can simulate believable human behavior. These agents remember and reflect days past as they plan for next days. This architecture extends LLM to store the complete record of the agents’ experiences and retrieve them dynamically to plan behavior. observation, planning, and reflection: three parts, each contribute critically to the believability of agent …

WWW 2023: Event prediction using case-based reasoning over knowledge graphs

Key idea: typical LP models are ill-suited for this task as they are incapable of performing inductive link prediction for new, unseen event entities and they require retraining as knowledge is added or changed in the underlying KG. This paper uses statistical measures to identify similar events and performs path-based predictions, requiring no training step. …

WWW 2023 KRACL: Contrastive Learning with Graph Context Modeling for Sparse Knowledge Graph Completion

Key idea: negative samples and loss functions Abstract Task: predict entities that appear less frequently in knowledge graphs. Contribution 1: propose knowledge relational attention network to leverage the graph context by simultaneously projecting neighboring triples to diferent latent spaces and jointly aggregating messages with the attention mechanism Contribution 2: propose knowledge contrastive loss by combining …

ACL 2022: SimpleKGC: Simple Contrastive Knowledge Graph Completion with Pre-trained Language ModelsPaper Documentation:

Abstract Task: Text-based methods still lag behind graph embedding-based methods like TransE, RotatE. They identify the key issue is efficient contrastive learning. Contribution 1: Introduce three types of negatives (more negatives than before): in-batch negatives, pre-batch negatives, and self negatives Contribution 2: Combined with InfoNCE loss, our model SimKGC outperforms other embedding-based methods on several …

Bert feature embedding of tweets

Pipeline Before feature engineering, use clean_tweet function to takes a raw tweet as input and removes URLs, user mentions, special characters, and extra spaces. The output is a cleaned tweet with only the relevant textual content. Then, use tokenizing function of pre-trained models to normalize the tweets. Then, get embeddings of each tweet using pre-trained …

Chatgpt Prompt Do and Don’t

Do not ask it to extract contents from a website, it is not connected to the internet. The content it returns is from its current dataset until 2021. If you want to ask it to help me write python code, it is better tell it which module you want it to import. The code quality …

Deploy Ethereum Full Node on AWS EC2

本地搭eth full node有点麻烦，不够serverless，加上之后也想和risingwave这个数据库打通，云端更方便。所以组里觉得我们还是deploy on AWS吧。然后搜了一下教程，https://pawelurbanek.com/ethereum-node-aws 是写的最好的，跟着步骤走下来大概一个半小时吧。 中间debug花了一点时间，遇到的问题是create a new file but no permission. 解决方法是用 chmod 777命令更改文件夹的权限。命令在下面 sudo chmod -R 777 /home/sixven/camp_sms/inputs 20/3/2023 更新 刚开始的时候还在问google能不能跑在aws s3上面，其实这个根本就不对。s3只是个存储。今天（一个礼拜之后）才想起来去问问chatgpt，算来浪费了一个小时。要拥抱gpt! 今后chatgpt要变成我的default option了。 然后是我给的这个 https://pawelurbanek.com/ethereum-node-aws 有问题的，因为这个没有加debug mode.后面再调整也不太可的样子。所以今天重新开了一个ec2 instance. 教训是需要去看官方文档，goethereum.org的东西，当时有点懒因为这个网站被nus给ban了所以没去上面double check. 偷一会儿懒后面就要用更多的时间去补了。