Large Language Models (LLMs) have shown impressive abilities in code generation, but they may generate erroneous programs. Reading a program takes ten times longer than writing it. Showing these erroneous programs to developers will waste developers' energies and introduce security risks to software. To address the above limitations, we propose HonestCoder, a novel LLM-based code generation approach. HonestCoder selectively shows the generated programs to developers based on LLMs' confidence. The confidence provides valuable insights into the correctness of generated programs. To achieve this goal, we propose a novel approach to estimate LLMs' confidence in code generation. It estimates confidence by measuring the multi-modal similarity between LLMs-generated programs. We collect and release a multilingual benchmark named TruthCodeBench, which consists of 2,265 samples and covers two popular programming languages (i.e., Python and Java). We apply HonestCoder to four popular LLMs (e.g., DeepSeek-Coder and Code Llama) and evaluate it on TruthCodeBench. Based on the experiments, we obtain the following insights. (1) HonestCoder can effectively estimate LLMs' confidence and accurately determine the correctness of generated programs. For example, HonestCoder outperforms the state-of-the-art baseline by 27.79% in AUROC and 63.74% in AUCPR. (2) HonestCoder can decrease the number of erroneous programs shown to developers. Compared to eight baselines, it can show more correct programs and fewer erroneous programs to developers. (3) Compared to showing code indiscriminately, HonestCoder only adds slight time overhead (approximately 0.4 seconds per requirement). (4) We discuss future directions to facilitate the application of LLMs in software development. We hope this work can motivate broad discussions about measuring the reliability of LLMs' outputs in performing code-related tasks.

Incorporating Euclidean symmetries (e.g. rotation equivariance) as inductive biases into graph neural networks has improved their generalization ability and data efficiency in unbounded physical dynamics modeling. However, in various scientific and engineering applications, the symmetries of dynamics are frequently discrete due to the boundary conditions. Thus, existing GNNs either overlook necessary symmetry, resulting in suboptimal representation ability, or impose excessive equivariance, which fails to generalize to unobserved symmetric dynamics. In this work, we propose a general Discrete Equivariant Graph Neural Network (DEGNN) that guarantees equivariance to a given discrete point group. Specifically, we show that such discrete equivariant message passing could be constructed by transforming geometric features into permutation-invariant embeddings. Through relaxing continuous equivariant constraints, DEGNN can employ more geometric feature combinations to approximate unobserved physical object interaction functions. Two implementation approaches of DEGNN are proposed based on ranking or pooling permutation-invariant functions. We apply DEGNN to various physical dynamics, ranging from particle, molecular, crowd to vehicle dynamics. In twenty scenarios, DEGNN significantly outperforms existing state-of-the-art approaches. Moreover, we show that DEGNN is data efficient, learning with less data, and can generalize across scenarios such as unobserved orientation.

Large Language Models (LLMs) have revolutionized the fields of computer vision (CV) and natural language processing (NLP). One of the most notable advancements of LLMs is that a single model is trained on vast and diverse datasets spanning multiple domains -- a paradigm we term `All in One'. This methodology empowers LLMs with super generalization capabilities, facilitating an encompassing comprehension of varied data distributions. Leveraging these capabilities, a single LLM demonstrates remarkable versatility across a variety of domains -- a paradigm we term `One for All'. However, applying this idea to the graph field remains a formidable challenge, with cross-domain pretraining often resulting in negative transfer. This issue is particularly important in few-shot learning scenarios, where the paucity of training data necessitates the incorporation of external knowledge sources. In response to this challenge, we propose a novel approach called Graph COordinators for PrEtraining (GCOPE), that harnesses the underlying commonalities across diverse graph datasets to enhance few-shot learning. Our novel methodology involves a unification framework that amalgamates disparate graph datasets during the pretraining phase to distill and transfer meaningful knowledge to target tasks. Extensive experiments across multiple graph datasets demonstrate the superior efficacy of our approach. By successfully leveraging the synergistic potential of multiple graph datasets for pretraining, our work stands as a pioneering contribution to the realm of graph foundational model.

Head of R&D of Google Cloud AI Jia Li has left her position with the company. Li informed Synced in a text message yesterday and the Google team confirmed her departure this morning. An Adjunct Professor at Stanford University's School of Medicine and a widely respected AI researcher, Li told Synced "I'm now pursuing the impact of AI for good in healthcare and working full-time at Stanford University's AIMI (Center for Artificial Intelligence in Medicine & Imaging). In healthcare, I am interested in how AI can improve the outcomes of individual patients as well as hospitals." Chinese media is reporting that Li will start her own AI company with the aim of bringing machine learning solutions to the healthcare industry; and that a number of leading global venture capitals are interested.

Yesterday, tech giant Google announced its latest solution, the Cloud AutoML, that will enable developers, even those that lack machine learning expertise, to build image recognition models. It is said to be a part of the company's initiative to democratize AI learning and provide a simple approach that anyone can easily understand. "Our goal was to lower the barrier of entry and make AI available to the largest possible community of developers, researchers and businesses," Fei-Fei Li, Google Cloud AI chief scientists, and Jia Li, Google Cloud AI Head of R&D, wrote in the company blog. According to the duo, their latest solution would help businesses with limited machine learning expertise build "their own high-quality custom models by using advanced techniques like learning2learn and transfer learning from Google." The two believe that Cloud AutoML will make experts in artificial intelligence more productive and take the technology to greater heights while helping less-skilled engineers build more powerful machine learning systems.

Reflecting the rapidly increasing interest and investment in cloud computing, 10,000 developers, engineers, IT executives, and Google employees and partners gathered at Next '17, Google's annual cloud event for enterprise customers. Google showcased customer testimonials from Disney, Verizon, HSBC, Colgate-Palmolive, and Ebay; support from a number of new partners, including SAP; and a series of AI and cloud infrastructure-related announcements. Analysts were not impressed (see here and here, for example). Google, the argument goes, is a consumer company and does not understand or have ready access to enterprise customers. Amazon's first mover advantage has produced an insurmountable market lead (last year, it reported cloud revenues of $12.22 billion, compared to Microsoft's $2.42 billion and Google's $900 million, Deutsche Bank estimates).

Google believes the key to growing its cloud computing business is artificial intelligence. The search giant said Tuesday that it had hired two high-profile AI researchers to lead a new machine learning unit that's part of its Google Cloud business. Machine learning is a subset of artificial intelligence that generally refers to training computers to recognize patterns amid tons of data. The two new hires are Fei-Fei Li, the director of Stanford University's Artificial Intelligence Lab; and Jia Li, the head of research for Snap, the parent company of popular social messaging app Snapchat. The two women are considered by analysts to be experts in the field of computer vision, a subset of artificial intelligence that involves teaching computers to recognize objects in images.

Today, Google Cloud chief Diane Greene announced the company's new push in machine learning and artificial intelligence. There's now a new group in Greene's division that will unify some of the disparate teams that had previously been doing machine learning work across Google's cloud. Two women will take charge of the new team: Fei-Fei Li, who was director of AI at Stanford, and Jia Li, who was previously head of research at Snap, Inc. As Business Insider notes, Jia Li was one of the minds behind the Snapchat feature that lets you attach emoji to real-world objects in your snaps. The news came at the top of a slew of announcements about the product roadmap for Google's cloud services and how they're expanding their use of machine learning, a critical technique for training large-scale AI networks to teach and improve themselves over time. The announcements were all aimed at showing how Google's cloud services include more than just renting time on a server -- that it can provide services to its enterprise customers that are based on its machine learning algorithms.