CNCast: Leveraging 3D Swin Transformer and DiT for Enhanced Regional Weather Forecasting

This study introduces a cutting-edge regional weather forecasting model based on the SwinTransformer 3D architecture. This model is specifically designed to deliver precise hourly weather predictions ranging from 1 hour to 5 days, significantly improving the reliability and practicality of short-term weather forecasts. Our model has demonstrated generally superior performance when compared to Pangu, a well-established global model. The evaluation indicates that our model excels in predicting most weather variables, highlighting its potential as a more effective alternative in the field of limited area modeling. A noteworthy feature of this model is the integration of enhanced boundary conditions, inspired by traditional numerical weather prediction (NWP) techniques. This integration has substantially improved the model's predictive accuracy. Additionally, the model includes an innovative approach for diagnosing hourly total precipitation at a high spatial resolution of approximately 5 kilometers. This is achieved through a latent diffusion model, offering an alternative method for generating high-resolution precipitation data.

Benchmarking and Understanding Compositional Relational Reasoning of LLMs

Compositional relational reasoning (CRR) is a hallmark of human intelligence, but we lack a clear understanding of whether and how existing transformer large language models (LLMs) can solve CRR tasks. To enable systematic exploration of the CRR capability of LLMs, we first propose a new synthetic benchmark called Generalized Associative Recall (GAR) by integrating and generalizing the essence of several tasks in mechanistic interpretability (MI) study in a unified framework. Evaluation shows that GAR is challenging enough for existing LLMs, revealing their fundamental deficiency in CRR. Meanwhile, it is easy enough for systematic MI study. Then, to understand how LLMs solve GAR tasks, we use attribution patching to discover the core circuits reused by Vicuna-33B across different tasks and a set of vital attention heads. Intervention experiments show that the correct functioning of these heads significantly impacts task performance. Especially, we identify two classes of heads whose activations represent the abstract notion of true and false in GAR tasks respectively. They play a fundamental role in CRR across various models and tasks. The dataset and code are available at https://github.com/Caiyun-AI/GAR.

Improving Transformers with Dynamically Composable Multi-Head Attention

Multi-Head Attention (MHA) is a key component of Transformer. In MHA, attention heads work independently, causing problems such as low-rank bottleneck of attention score matrices and head redundancy. We propose Dynamically Composable Multi-Head Attention (DCMHA), a parameter and computation efficient attention architecture that tackles the shortcomings of MHA and increases the expressive power of the model by dynamically composing attention heads. At the core of DCMHA is a $\it{Compose}$ function that transforms the attention score and weight matrices in an input-dependent way. DCMHA can be used as a drop-in replacement of MHA in any transformer architecture to obtain the corresponding DCFormer. DCFormer significantly outperforms Transformer on different architectures and model scales in language modeling, matching the performance of models with ~1.7x-2.0x compute. For example, DCPythia-6.9B outperforms open source Pythia-12B on both pretraining perplexity and downstream task evaluation. The code and models are available at https://github.com/Caiyun-AI/DCFormer.

InterHT: Knowledge Graph Embeddings by Interaction between Head and Tail Entities

Knowledge graph embedding (KGE) models learn the representation of entities and relations in knowledge graphs. Distance-based methods show promising performance on link prediction task, which predicts the result by the distance between two entity representations. However, most of these methods represent the head entity and tail entity separately, which limits the model capacity. We propose a novel distance-based method named InterHT that allows the head and tail entities to interact better and get better entity representation. Experimental results show that our proposed method achieves the best results on ogbl-wikikg2 dataset.

PM2.5-GNN: A Domain Knowledge Enhanced Graph Neural Network For PM2.5 Forecasting

When predicting PM2.5 concentrations, it is necessary to consider complex information sources since the concentrations are influenced by various factors within a long period. In this paper, we identify a set of critical domain knowledge for PM2.5 forecasting and develop a novel graph based model, PM2.5-GNN, being capable of capturing long-term dependencies. On a real-world dataset, we validate the effectiveness of the proposed model and examine its abilities of capturing both fine-grained and long-term influences in PM2.5 process. The proposed PM2.5-GNN has also been deployed online to provide free forecasting service.