Wenyi Hu

Yao Shu, Wenyang Hu, See-Kiong Ng et al.

Large Language Models (LLMs) have become indispensable in numerous real-world applications. However, fine-tuning these models at scale, especially in federated settings where data privacy and communication efficiency are critical, presents significant challenges. Existing approaches often resort to parameter-efficient fine-tuning (PEFT) to mitigate communication overhead, but this typically comes at the cost of model accuracy. To this end, we propose federated full-parameter tuning at scale for LLMs (Ferret), the first first-order method with shared randomness to enable scalable full-parameter tuning of LLMs across decentralized data sources while maintaining competitive model accuracy. Ferret accomplishes this through three aspects: (i) it employs widely used first-order methods for efficient local updates; (ii) it projects these updates into a low-dimensional space to considerably reduce communication overhead; and (iii) it reconstructs local updates from this low-dimensional space with shared randomness to facilitate effective full-parameter global aggregation, ensuring fast convergence and competitive final performance. Our rigorous theoretical analyses and insights along with extensive experiments, show that Ferret significantly enhances the scalability of existing federated full-parameter tuning approaches by achieving high computational efficiency, reduced communication overhead, and fast convergence, all while maintaining competitive model accuracy. Our implementation is available at https://github.com/allen4747/Ferret.

Gregory Kang Ruey Lau, Wenyang Hu, Diwen Liu et al.

Large Language Models (LLMs), particularly smaller variants, still struggle with complex reasoning tasks. While inference-time prompting can guide reasoning, existing methods often rely on sequential queries. Ensemble approaches offer a promising path to performance gains, especially given recent batch inference speed-ups. This work introduces DIPPER, a novel, training-free framework that transforms a single LLM into an effective inference-time ensemble. By feeding the model an optimized and diverse set of prompts in parallel, DIPPER elicits varied reasoning paths, leading to performance gains. We empirically demonstrate significant improvements on reasoning benchmarks, such as MATH, where a DIPPER ensemble of three Qwen2-MATH-1.5B instances (via parallel prompting of a single model) outperforms a larger 7B model.

Arun Verma, Zhaoxuan Wu, Zijian Zhou et al.

Large Language Models (LLMs) are large-scale pretrained models that have achieved remarkable success across diverse domains. These successes have been driven by unprecedented complexity and scale in both data and computations. However, due to the high costs of training such models, brute-force trial-and-error approaches to improve LLMs are not feasible. Inspired by the success of inverse problems in uncovering fundamental scientific laws, this position paper advocates that inverse problems can also efficiently uncover scaling laws that guide the building of LLMs to achieve the desirable performance with significantly better cost-effectiveness.

Pengyu Yuan, Wenyi Hu, Xuqing Wu et al.

We proposed a robust segmentation and picking workflow to solve the first arrival picking problem for seismic signal processing. Unlike traditional classification algorithm, image segmentation method can utilize the location information by outputting a prediction map which has the same size of the input image. A parameter-free nearest point picking algorithm is proposed to further improve the accuracy of the first arrival picking. The algorithm is test on synthetic clean data, synthetic noisy data, synthetic picking-disconnected data and field data. It performs well on all of them and the picking deviation reaches as low as 4.8ms per receiver. The first arrival picking problem is formulated as the contour detection problem. Similar to \cite{wu2019semi}, we use U-net to perform the segmentation as it is proven to be state-of-the-art in many image segmentation tasks. Particularly, a Lovasz loss instead of the traditional cross-entropy loss is used to train the network for a better segmentation performance. Lovasz loss is a surrogate loss for Jaccard index or the so-called intersection-over-union (IoU) score, which is often one of the most used metrics for segmentation tasks. In the picking part, we use a novel nearest point picking (NPP) method to take the advantage of the coherence of the first arrival picking among adjacent receivers. Our model is tested and validated on both synthetic and field data with harmonic noises. The main contributions of this paper are as follows: 1. Used Lovasz loss to directly optimize the IoU for segmentation task. Improvement over the cross-entropy loss with regard to the segmentation accuracy is verified by the test result. 2. Proposed a nearest point picking post processing method to overcome any defects left by the segmentation output. 3. Conducted noise analysis and verified the model with both noisy synthetic and field datasets.

Wenyi Hu, Yuchen Jin, Xuqing Wu et al.

For the purpose of effective suppression of the cycle-skipping phenomenon in full waveform inversion (FWI), we developed a Deep Neural Network (DNN) approach to predict the absent low-frequency components by exploiting the implicit relation connecting the low-frequency and high-frequency data through the subsurface geological and geophysical properties. In order to solve this challenging nonlinear regression problem, two novel strategies were proposed to design the DNN architecture and the learning workflow: 1) Dual Data Feed; 2) Progressive Transfer Learning. With the Dual Data Feed structure, both the high-frequency data and the corresponding Beat Tone data are fed into the DNN to relieve the burden of feature extraction, thus reducing the network complexity and the training cost. The second strategy, Progressive Transfer Learning, enables us to unbiasedly train the DNN using a single training dataset. Unlike most established deep learning approaches where the training datasets are fixed, within the framework of the Progressive Transfer Learning, the training dataset evolves in an iterative manner while gradually absorbing the subsurface information retrieved by the physics-based inversion module, progressively enhancing the prediction accuracy of the DNN and propelling the FWI process out of the local minima. The Progressive Transfer Learning, alternatingly updating the training velocity model and the DNN parameters in a complementary fashion toward convergence, saves us from being overwhelmed by the otherwise tremendous amount of training data, and avoids the underfitting and biased sampling issues. The numerical experiments validated that, without any a priori geological information, the low-frequency data predicted by the Progressive Transfer Learning are sufficiently accurate for an FWI engine to produce reliable subsurface velocity models free of cycle-skipping-induced artifacts.