Quantifying the impact of training data points is crucial for understanding the outputs of machine learning models and for improving the transparency of the AI pipeline. The influence function is a principled and popular data attribution method, but its computational cost often makes it challenging to use. This issue becomes more pronounced in the setting of large language models and text-to-image models. In this work, we propose DataInf, an efficient influence approximation method that is practical for large-scale generative AI models. Leveraging an easy-to-compute closed-form expression, DataInf outperforms existing influence computation algorithms in terms of computational and memory efficiency. Our theoretical analysis shows that DataInf is particularly well-suited for parameter-efficient fine-tuning techniques such as LoRA. Through systematic empirical evaluations, we show that DataInf accurately approximates influence scores and is orders of magnitude faster than existing methods. In applications to RoBERTa-large, Llama-2-13B-chat, and stable-diffusion-v1.5 models, DataInf effectively identifies the most influential fine-tuning examples better than other approximate influence scores. Moreover, it can help to identify which data points are mislabeled.

Retrieval augmented generation (RAG) is frequently used to mitigate hallucinations and provide up-to-date knowledge for large language models (LLMs). However, given that document retrieval is an imprecise task and sometimes results in erroneous or even harmful content being presented in context, this raises the question of how LLMs handle retrieved information: If the provided content is incorrect, does the model know to ignore it, or does it recapitulate the error? Conversely, when the model's initial response is incorrect, does it always know to use the retrieved information to correct itself, or does it insist on its wrong prior response? To answer this, we curate a dataset of over 1200 questions across six domains (e.g., drug dosages, Olympic records, locations) along with content relevant to answering each question. We further apply precise perturbations to the answers in the content that range from subtle to blatant errors.We benchmark six top-performing LLMs, including GPT-4o, on this dataset and find that LLMs are susceptible to adopting incorrect retrieved content, overriding their own correct prior knowledge over 60\% of the time. However, the more unrealistic the retrieved content is (i.e. more deviated from truth), the less likely the model is to adopt it. Also, the less confident a model is in its initial response (via measuring token probabilities), the more likely it is to adopt the information in the retrieved content. We exploit this finding and demonstrate simple methods for improving model accuracy where there is conflicting retrieved content. Our results highlight a difficult task and benchmark for LLMs -- namely, their ability to correctly discern when it is wrong in light of correct retrieved content and to reject cases when the provided content is incorrect. Our dataset, called ClashEval, and evaluations are open-sourced to allow for future benchmarking on top-performing models at https://github.com/kevinwu23/StanfordClashEval.

OsciBear is a system-on-chip (SoC) featuring a RISC-V 32-bit 5-stage in-order scalar processor, AES accelerator, BLE 1M baseband-modem, and a 2.4 GHz radio front end (RFE) transceiver. It was designed in TSMC's 28nm process with a total die area of 1 mm2during the course of a 14-week semester by 18 students - 4 Ph.D students, 6 masters students, and 8 undergraduates - enrolled in UC Berkeley's special topics course “28nm SoC for loT” in Spring 2021. Additionally, a PCB was designed with off-chip reference clocks, bring-up tooling, as well as power amplifiers, RF switch, and an antenna to complete the radio front-end. The CPU has been demonstrated to run up to 30 MHz in typical operating conditions. The BLE 1M-compliant PHY layer packet assembly and disassembly has been verified in-hardware through “loopback” testing. Adherence to BLE's PHY FM specifications has also been verified with a commercial BLE receiver. In total, the chip consumes 8.43 mW of static power.

Visual information arriving at the retina is transmitted to the brain by signals in the optic nerve, and the brain must rely solely on these signals to make inferences about the visual world. Previous work has probed the content of these signals by directly reconstructing images from retinal activity using linear regression or nonlinear regression with neural networks. Maximum a posteriori (MAP) reconstruction using retinal encoding models and separately-trained natural image priors offers a more general and principled approach. We develop a novel method for approximate MAP reconstruction that combines a generalized linear model for retinal responses to light, including their dependence on spike history and spikes of neighboring cells, with the image prior implicitly embedded in a deep convolutional neural network trained for image denoising. We use this method to reconstruct natural images from ex vivo simultaneously-recorded spikes of hundreds of retinal ganglion cells uniformly sampling a region of the retina. The method produces reconstructions that match or exceed the state-of-the-art in perceptual similarity and exhibit additional fine detail, while using substantially fewer model parameters than previous approaches. The use of more rudimentary encoding models (a linear-nonlinear-Poisson cascade) or image priors (a 1/f spectral model) significantly reduces reconstruction performance, indicating the essential role of both components in achieving high-quality reconstructed images from the retinal signal.

SUMMARY: RNA sequencing has provided unprecedented resolution of alternative splicing and splicing quantitative trait loci (sQTL). However, there are few tools available for visualizing the genotype-dependent effects of splicing at a population level. SplicePlot is a simple command line utility that produces intuitive visualization of sQTLs and their effects. SplicePlot takes mapped RNA sequencing reads in BAM format and genotype data in VCF format as input and outputs publication-quality Sashimi plots, hive plots and structure plots, enabling better investigation and understanding of the role of genetics on alternative splicing and transcript structure. AVAILABILITY AND IMPLEMENTATION: Source code and detailed documentation are available at http://montgomerylab.stanford.edu/spliceplot/index.html under Resources and at Github. SplicePlot is implemented in Python and is supported on Linux and Mac OS. A VirtualBox virtual machine running Ubuntu with SplicePlot already installed is also available.

Sleep is a basic physiological process essential for good health. However, 40 million people in the U.S. are diagnosed with sleep disorders, with many more undiagnosed. To help address this problem, we developed an application, ShutEye, which provides a peripheral display on the wall-paper of the user's mobile phone to promote awareness about recommended activities that promote good sleep quality. Based on preferences about the user's desired bed-time and activities' for example - consuming caffeine or performing vigorous exercise - ShutEye displays guidance about when engaging in those activities is likely to affect sleep without requiring any explicit interaction from the user. In this paper, we describe ShutEye and results from a four-week field study with 12 participants. Results indicate that a simple, recommendation-based peripheral display can be a very low-effort but still effective method for improving awareness of healthy sleep habits. We also provide recommendations about designing peripheral displays and extend insights for designing health-based mobile applications.