Add to Chrome
Add to Firefox
Add to EdgeTuan Dung Nguyen
AstroLLaMA-Chat: Scaling AstroLLaMA with Conversational and Diverse Datasets
Jan 05, 2024
We explore the potential of enhancing LLM performance in astronomy-focused question-answering through targeted, continual pre-training. By employing a compact 7B-parameter LLaMA-2 model and focusing exclusively on a curated set of astronomy corpora -- comprising abstracts, introductions, and conclusions -- we achieve notable improvements in specialized topic comprehension. While general LLMs like GPT-4 excel in broader question-answering scenarios due to superior reasoning capabilities, our findings suggest that continual pre-training with limited resources can still enhance model performance on specialized topics. Additionally, we present an extension of AstroLLaMA: the fine-tuning of the 7B LLaMA model on a domain-specific conversational dataset, culminating in the release of the chat-enabled AstroLLaMA for community use. Comprehensive quantitative benchmarking is currently in progress and will be detailed in an upcoming full paper. The model, AstroLLaMA-Chat, is now available at https://huggingface.co/universeTBD, providing the first open-source conversational AI tool tailored for the astronomy community.
On Partial Optimal Transport: Revising the Infeasibility of Sinkhorn and Efficient Gradient Methods
Dec 22, 2023This paper studies the Partial Optimal Transport (POT) problem between two unbalanced measures with at most $n$ supports and its applications in various AI tasks such as color transfer or domain adaptation. There is hence the need for fast approximations of POT with increasingly large problem sizes in arising applications. We first theoretically and experimentally investigate the infeasibility of the state-of-the-art Sinkhorn algorithm for POT due to its incompatible rounding procedure, which consequently degrades its qualitative performance in real world applications like point-cloud registration. To this end, we propose a novel rounding algorithm for POT, and then provide a feasible Sinkhorn procedure with a revised computation complexity of $\mathcal{\widetilde O}(n^2/\varepsilon^4)$. Our rounding algorithm also permits the development of two first-order methods to approximate the POT problem. The first algorithm, Adaptive Primal-Dual Accelerated Gradient Descent (APDAGD), finds an $\varepsilon$-approximate solution to the POT problem in $\mathcal{\widetilde O}(n^{2.5}/\varepsilon)$, which is better in $\varepsilon$ than revised Sinkhorn. The second method, Dual Extrapolation, achieves the computation complexity of $\mathcal{\widetilde O}(n^2/\varepsilon)$, thereby being the best in the literature. We further demonstrate the flexibility of POT compared to standard OT as well as the practicality of our algorithms on real applications where two marginal distributions are unbalanced.
Federated Deep Equilibrium Learning: A Compact Shared Representation for Edge Communication Efficiency
Sep 27, 2023
Federated Learning (FL) is a prominent distributed learning paradigm facilitating collaboration among nodes within an edge network to co-train a global model without centralizing data. By shifting computation to the network edge, FL offers robust and responsive edge-AI solutions and enhance privacy-preservation. However, deploying deep FL models within edge environments is often hindered by communication bottlenecks, data heterogeneity, and memory limitations. To address these challenges jointly, we introduce FeDEQ, a pioneering FL framework that effectively employs deep equilibrium learning and consensus optimization to exploit a compact shared data representation across edge nodes, allowing the derivation of personalized models specific to each node. We delve into a unique model structure composed of an equilibrium layer followed by traditional neural network layers. Here, the equilibrium layer functions as a global feature representation that edge nodes can adapt to personalize their local layers. Capitalizing on FeDEQ's compactness and representation power, we present a novel distributed algorithm rooted in the alternating direction method of multipliers (ADMM) consensus optimization and theoretically establish its convergence for smooth objectives. Experiments across various benchmarks demonstrate that FeDEQ achieves performance comparable to state-of-the-art personalized methods while employing models of up to 4 times smaller in communication size and 1.5 times lower memory footprint during training.
AstroLLaMA: Towards Specialized Foundation Models in Astronomy
Sep 12, 2023
Large language models excel in many human-language tasks but often falter in highly specialized domains like scholarly astronomy. To bridge this gap, we introduce AstroLLaMA, a 7-billion-parameter model fine-tuned from LLaMA-2 using over 300,000 astronomy abstracts from arXiv. Optimized for traditional causal language modeling, AstroLLaMA achieves a 30% lower perplexity than Llama-2, showing marked domain adaptation. Our model generates more insightful and scientifically relevant text completions and embedding extraction than state-of-the-arts foundation models despite having significantly fewer parameters. AstroLLaMA serves as a robust, domain-specific model with broad fine-tuning potential. Its public release aims to spur astronomy-focused research, including automatic paper summarization and conversational agent development.
On the Generalization of Wasserstein Robust Federated Learning
Jun 03, 2022
In federated learning, participating clients typically possess non-i.i.d. data, posing a significant challenge to generalization to unseen distributions. To address this, we propose a Wasserstein distributionally robust optimization scheme called WAFL. Leveraging its duality, we frame WAFL as an empirical surrogate risk minimization problem, and solve it using a local SGD-based algorithm with convergence guarantees. We show that the robustness of WAFL is more general than related approaches, and the generalization bound is robust to all adversarial distributions inside the Wasserstein ball (ambiguity set). Since the center location and radius of the Wasserstein ball can be suitably modified, WAFL shows its applicability not only in robustness but also in domain adaptation. Through empirical evaluation, we demonstrate that WAFL generalizes better than the vanilla FedAvg in non-i.i.d. settings, and is more robust than other related methods in distribution shift settings. Further, using benchmark datasets we show that WAFL is capable of generalizing to unseen target domains.
DONE: Distributed Approximate Newton-type Method for Federated Edge Learning
Jan 17, 2021
There is growing interest in applying distributed machine learning to edge computing, forming federated edge learning. Federated edge learning faces non-i.i.d and heterogeneous data, and the communication between edge workers, possibly through distant locations and with unstable wireless networks, is more costly than their local computational overhead. Here, we propose DONE, a distributed approximate Newton-type algorithm with fast convergence rate for communication-efficient federated edge learning. First, with strongly convex and smooth loss functions, DONE can approximately produce the Newton direction in a distributed manner by using the classical Richardson iteration on each edge worker. Second, we prove that DONE has linear-quadratic convergence and analyze its computation and communication complexities. Finally, the experimental results with non-i.i.d. and heterogeneous data show that DONE attains comparable performance to the Newton's method. Notably, DONE requires fewer communication iterations compared to distributed gradient descent and outperforms DANE, a similar and state-of-the-art approach, in the case of non-quadratic loss functions.
Personalized Federated Learning with Moreau Envelopes
Jun 16, 2020
Federated learning (FL) is a decentralized and privacy-preserving machine learning technique in which a group of clients collaborate with a server to learn a global model without sharing clients' data. One challenge associated with FL is statistical diversity among clients, which restricts the global model from delivering good performance on each client's task. To address this, we propose an algorithm for personalized FL (pFedMe) using Moreau envelopes as clients' regularized loss functions, which help decouple personalized model optimization from the global model learning in a bi-level problem stylized for personalized FL. Theoretically, we show that pFedMe's convergence rate is state-of-the-art: achieving quadratic speedup for strongly convex and sublinear speedup of order 2/3 for smooth nonconvex objectives. Experimentally, we verify that pFedMe excels at empirical performance compared with the vanilla FedAvg and Per-FedAvg, a meta-learning based personalized FL algorithm.