Augmenting LLMs with context leads to improved performance across many applications. Despite much research on Retrieval Augmented Generation (RAG) systems, an open question is whether errors arise because LLMs fail to utilize the context from retrieval or the context itself is insufficient to answer the query. To shed light on this, we develop a new notion of sufficient context, along with a method to classify instances that have enough information to answer the query. We then use sufficient context to analyze several models and datasets. By stratifying errors based on context sufficiency, we find that larger models with higher baseline performance (Gemini 1.5 Pro, GPT 4o, Claude 3.5) excel at answering queries when the context is sufficient, but often output incorrect answers instead of abstaining when the context is not. On the other hand, smaller models with lower baseline performance (Llama 3.1, Mistral 3, Gemma 2) hallucinate or abstain often, even with sufficient context. We further categorize cases when the context is useful, and improves accuracy, even though it does not fully answer the query and the model errs without the context. Building on our findings, we explore ways to reduce hallucinations in RAG systems, including a new selective generation method that leverages sufficient context information for guided abstention. Our method improves the fraction of correct answers among times where the model responds by 2--10% for Gemini, GPT, and Gemma. Code for our selective generation method and the prompts used in our autorater analysis are available on our [github](https://github.com/hljoren/sufficientcontext).

Large language models (LLMs) have demonstrated remarkable capabilities, but their outputs can sometimes be unreliable or factually incorrect. To address this, we introduce Self Logits Evolution Decoding (SLED), a novel decoding framework that enhances the truthfulness of LLMs without relying on external knowledge bases or requiring further fine-tuning. From an optimization perspective, our SLED framework leverages the latent knowledge embedded within the LLM by contrasting the output logits from the final layer with those from early layers. It then utilizes an approximate gradient approach to enable latent knowledge to guide the self-refinement of outputs, thereby effectively improving factual accuracy. Extensive experiments have been conducted on established benchmarks across a diverse range of model families (LLaMA 2, LLaMA 3, Gemma) and scales (from 2B to 70B), including more advanced architectural configurations such as the mixture of experts (MoE). Our evaluation spans a wide variety of tasks, including multi-choice, open-generation, and adaptations to chain-of-thought reasoning tasks. The results demonstrate that SLED consistently improves factual accuracy by up to 20\% compared to existing decoding methods while maintaining natural language fluency and negligible latency overhead. Furthermore, it can be flexibly combined with other decoding methods to further enhance their performance.

Adversarial robustness corresponds to the susceptibility of deep neural networks to imperceptible perturbations made at test time. In the context of image tasks, many algorithms have been proposed to make neural networks robust to adversarial perturbations made to the input pixels. These perturbations are typically measured in an ℓpnorm. However, robustness often holds only for the specific attack used for training. In this work we extend the above setting to consider the problem of training of deep neural networks that can be made simultaneously robust to perturbations applied in multiple natural representations spaces. For the case of image data, examples include the standard pixel representation as well as the representation in the discrete cosine transform (DCT) basis. We design a theoretically sound algorithm with formal guarantees for the above problem. Furthermore, our guarantees also hold when the goal is to require robustness with respect to multiple ℓpnorm based attacks. We then derive an efficient practical implementation and demonstrate the effectiveness of our approach on standard datasets for image classification.1

Transformer models are permutation equivariant.To supply the order and type information of the input tokens, position and segment embeddings are usually added to the input.Recent works proposed variations of positional encodings with relative position encodings achieving better performance.Our analysis shows that the gain actually comes from moving positional information to attention layer from the input.Motivated by this, we introduce Decoupled posItional attEntion for Transformers (DIET), a simple yet effective mechanism to encode position and segment information into the Transformer models.The proposed method has faster training and inference time, while achieving competitive performance on GLUE, XTREME and WMT benchmarks.We further generalize our method to long-range transformers and show performance gain.

We present Neural Structured Learning (NSL) in TensorFlow [1], a new learning paradigm to train neural networks by leveraging structured signals in addition to feature inputs. Structure can be explicit as represented by a graph, or implicit, either induced by adversarial perturbation or inferred using techniques like embedding learning. NSL is open-sourced as part of the TensorFlow [2] ecosystem and is widely used in Google across many products and services. In this tutorial, we provide an overview of the NSL framework including various libraries, tools, and APIs as well as demonstrate the practical use of NSL in different applications. The NSL website is hosted at www.tensorflow.org/neural_structured_learning, which includes details about the theoretical foundations of the technology, extensive API documentation, and hands-on tutorials.

We present Neural Structured Learning (NSL) in TensorFlow [2], a new learning paradigm to train neural networks by leveraging structured signals in addition to feature inputs. Structure can be explicit as represented by a graph, or implicit, either induced by adversarial perturbation or inferred using techniques like embedding learning. NSL is open-sourced as part of the TensorFlow [3] ecosystem and is widely used in Google across many products and services. In this tutorial, we provide an overview of the NSL framework including various libraries, tools, and APIs as well as demonstrate the practical use of NSL in different applications. The NSL website is hosted at www.tensorflow.org/neural_structured_learning, which includes details about the theoretical foundations of the technology, extensive API documentation, and hands-on tutorials.

The abundance of real-world data and limited labeling budget calls for active learning, an important learning paradigm for reducing human labeling efforts. Many recently developed active learning algorithms consider both uncertainty and representativeness when making querying decisions. However, exploiting representativeness with uncertainty concurrently usually requires tackling sophisticated and challenging learning tasks, such as clustering. In this letter, we propose a new active learning framework, called hinted sampling, which takes both uncertainty and representativeness into account in a simpler way. We design a novel active learning algorithm within the hinted sampling framework with an extended support vector machine. Experimental results validate that the novel active learning algorithm can result in a better and more stable performance than that achieved by state-of-the-art algorithms. We also show that the hinted sampling framework allows improving another active learning algorithm designed from the transductive support vector machine.

We formulate a framework for applying error-correcting codes (ECCs) on multilabel classification problems. The framework treats some base learners as noisy channels and uses ECC to correct the prediction errors made by the learners. The framework immediately leads to a novel ECC-based explanation of the popular random k-label sets (RAKEL) algorithm using a simple repetition ECC. With the framework, we empirically compare a broad spectrum of off-the-shelf ECC designs for multilabel classification. The results not only demonstrate that RAKEL can be improved by applying some stronger ECC, but also show that the traditional binary relevance approach can be enhanced by learning more parity-checking labels. Our research on different ECCs also helps to understand the tradeoff between the strength of ECC and the hardness of the base learning tasks. Furthermore, we extend our research to ECC with either hard (binary) or soft (real-valued) bits by designing a novel decoder. We demonstrate that the decoder improves the performance of our framework.

Efficient image search clustering is prominent for image search engines for exponentially growing photo collections. In this work, we propose an image search clustering approach which selects multiple canonical images from image search results and constructs image clusters in real time on an image sub-graph for the search results. The efficiency is achieved with the help of offline-computed image context graphs by distributed computing methods. Extending our prior works, we demonstrate the results of the proposed canonical image selection and preliminary outcomes of large-scale image graph construction in this proposal. We experiment in Flickr550 dataset, containing 540,321 Flickr photos.