arXiv subjects

Yao Fu

Publications and source records attributed to Yao Fu.

At least 73 records · Page 4Linked to original sources

arXiv2023

Breaking Rate-Distance Limitation of Measurement-Device-Independent Quantum Secret Sharing

Currently most progresses on quantum secret sharing suffer from rate-distance bound, and thus the key rates are limited. In addition to the limited key rate, the technical difficulty and the corresponding cost together prevent large-scale deployment. Furthermore, the performance of most existing protocols is analyzed in the asymptotic regime without considering participant attacks. Here we report a measurement-device-independent quantum secret sharing protocol with improved key rate and transmission distance. Based on spatial multiplexing, our protocol shows it can break rate-distance bounds over network under at least ten communication parties. Compared with other protocols, our work improves the secret key rate by more than two orders of magnitude and has a longer transmission distance. We analyze the security of our protocol in the composable framework considering participant attacks and evaluate its performance in the finite-size regime. In addition, we investigate applying our protocol to digital signatures where the signature rate is improved more than $10^7$ times compared with existing protocols. We anticipate that our quantum secret sharing protocol will provide a solid future for multiparty applications on the quantum network.

quant-ph↗

arXiv2023

Asynchronous measurement-device-independent quantum key distribution with hybrid source

The linear constraint of secret key rate capacity is overcome by the tiwn-field quantum key distribution (QKD). However, the complex phase-locking and phase-tracking technique requirements throttle the real-life applications of twin-field protocol. The asynchronous measurement-device-independent (AMDI) QKD or called mode-pairing QKD protocol can relax the technical requirements and keep the similar performance of twin-field protocol. Here, we propose an AMDI-QKD protocol with a nonclassical light source by changing the phase-randomized weak coherent state to a phase-randomized coherent-state superposition in the signal state time window. Simulation results show that our proposed hybrid source protocol significantly enhances the key rate of the AMDI-QKD protocol, while exhibiting robustness to imperfect modulation of nonclassical light sources.

quant-ph↗

arXiv2023

Probing Parton distribution functions at large x via Drell-Yan Forward-Backward Asymmetry

The forward-backward asymmetry of the Drell-Yan process in dilepton decays at high invariant masses can be used to probe the parton distribution functions at large x. The behavior of three modern PDF sets (CT18NNLO, MSHT20, and NNPDF4.0) are compared, and updated under various scenarios via ePump using proton-proton collision pseudo-data generated at sqrt(s) = 13 TeV with 3000 fb^{-1} of integrated luminosity.

hep-ph↗

arXiv2023

Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performance

As large language models (LLMs) are continuously being developed, their evaluation becomes increasingly important yet challenging. This work proposes Chain-of-Thought Hub, an open-source evaluation suite on the multi-step reasoning capabilities of large language models. We are interested in this setting for two reasons: (1) from the behavior of GPT and PaLM model family, we observe that complex reasoning is likely to be a key differentiator between weaker and stronger LLMs; (2) we envisage large language models to become the next-generation computational platform and foster an ecosystem of LLM-based new applications, this naturally requires the foundation models to perform complex tasks that often involve the composition of linguistic and logical operations. Our approach is to compile a suite of challenging reasoning benchmarks to track the progress of LLMs. Our current results show that: (1) model scale clearly correlates with reasoning capabilities; (2) As of May 2023, Claude-v1.3 and PaLM-2 are the only two models that are comparable with GPT-4, while open-sourced models still lag behind; (3) LLaMA-65B performs closely to code-davinci-002, indicating that with successful further development such as reinforcement learning from human feedback (RLHF), it has great potential to be close to GPT-3.5-Turbo. Our results also suggest that for the open-source efforts to catch up, the community may focus more on building better base models and exploring RLHF.

cs.CL↗

arXiv2023

Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedback

We study whether multiple large language models (LLMs) can autonomously improve each other in a negotiation game by playing, reflecting, and criticizing. We are interested in this question because if LLMs were able to improve each other, it would imply the possibility of creating strong AI agents with minimal human intervention. We ask two LLMs to negotiate with each other, playing the roles of a buyer and a seller, respectively. They aim to reach a deal with the buyer targeting a lower price and the seller a higher one. A third language model, playing the critic, provides feedback to a player to improve the player's negotiation strategies. We let the two agents play multiple rounds, using previous negotiation history and AI feedback as in-context demonstrations to improve the model's negotiation strategy iteratively. We use different LLMs (GPT and Claude) for different roles and use the deal price as the evaluation metric. Our experiments reveal multiple intriguing findings: (1) Only a subset of the language models we consider can self-play and improve the deal price from AI feedback, weaker models either do not understand the game's rules or cannot incorporate AI feedback for further improvement. (2) Models' abilities to learn from the feedback differ when playing different roles. For example, it is harder for Claude-instant to improve as the buyer than as the seller. (3) When unrolling the game to multiple rounds, stronger agents can consistently improve their performance by meaningfully using previous experiences and iterative AI feedback, yet have a higher risk of breaking the deal. We hope our work provides insightful initial explorations of having models autonomously improve each other with game playing and AI feedback.

cs.CL↗

arXiv2023

Decomposed Prompting: A Modular Approach for Solving Complex Tasks

Few-shot prompting is a surprisingly powerful way to use Large Language Models (LLMs) to solve various tasks. However, this approach struggles as the task complexity increases or when the individual reasoning steps of the task themselves are hard to learn, especially when embedded in more complex tasks. To address this, we propose Decomposed Prompting, a new approach to solve complex tasks by decomposing them (via prompting) into simpler sub-tasks that can be delegated to a library of prompting-based LLMs dedicated to these sub-tasks. This modular structure allows each prompt to be optimized for its specific sub-task, further decomposed if necessary, and even easily replaced with more effective prompts, trained models, or symbolic functions if desired. We show that the flexibility and modularity of Decomposed Prompting allows it to outperform prior work on few-shot prompting using GPT3. On symbolic reasoning tasks, we can further decompose sub-tasks that are hard for LLMs into even simpler solvable sub-tasks. When the complexity comes from the input length, we can recursively decompose the task into the same task but with smaller inputs. We also evaluate our approach on textual multi-step reasoning tasks: on long-context multi-hop QA task, we can more effectively teach the sub-tasks via our separate sub-tasks prompts; and on open-domain multi-hop QA, we can incorporate a symbolic information retrieval within our decomposition framework, leading to improved performance on both tasks. Datasets, Code and Prompts available at https://github.com/allenai/DecomP.

cs.CL↗

arXiv2023

Breaking universal limitations on quantum conference key agreement without quantum memory

Quantum conference key agreement is an important cryptographic primitive for future quantum network. Realizing this primitive requires high-brightness and robust multiphoton entanglement sources, which is challenging in experiment and unpractical in application because of limited transmission distance caused by channel loss. Here we report a measurement-device-independent quantum conference key agreement protocol with enhanced transmission efficiency over lossy channel. With spatial multiplexing nature and adaptive operation, our protocol can break key rate bounds on quantum communication over quantum network without quantum memory. Compared with previous work, our protocol shows superiority in key rate and transmission distance within the state-of-the-art technology. Furthermore, we analyse the security of our protocol in the composable framework and evaluate its performance in the finite-size regime to show practicality. Based on our results, we anticipate that our protocol will play an indispensable role in constructing multipartite quantum network.

quant-ph↗

arXiv2023

Scalable High-Rate Twin-Field Quantum Key Distribution Networks without Constraint of Probability and Intensity

Implementation of a twin-field quantum key distribution network faces limitations, including the low tolerance of interference errors for phase-matching type protocols and the strict constraint regarding intensity and probability for sending-or-not-sending type protocols. Here, we propose a two-photon twin-field quantum key distribution protocol and achieve twin-field-type two-photon interference through post-matching phase-correlated single-photon interference events. We exploit the non-interference mode as the code mode to highly tolerate interference errors, and the two-photon interference naturally removes the intensity and probability constraint. Therefore, our protocol can transcend the abovementioned limitations while breaking the secret key capacity of repeaterless quantum key distribution. Simulations show that for a four-user networks, under which each node with fixed system parameters can dynamically switch different attenuation links, the key rates of our protocol for all six links can either exceed or approach the secret key capacity. However, the key rates of all links are lower than the key capacity when using phase-matching type protocols. Additionally, four of the links could not extract the key when using sending-or-not-sending type protocols. We anticipate that our protocol can facilitate the development of practical and efficient quantum networks.

quant-ph↗

arXiv2023

Experimental quantum secret sharing based on phase encoding of coherent states

Quantum secret sharing (QSS) is one of the basic communication primitives in future quantum networks which addresses part of the basic cryptographic tasks of multiparty communication and computation. Nevertheless, it is a challenge to provide a practical QSS protocol with security against general attacks. A QSS protocol that balances security and practicality is still lacking. Here, we propose a QSS protocol with simple phase encoding of coherent states among three parties. Removing the requirement of impractical entangled resources and the need for phase randomization, our protocol can be implemented with accessible technology. We provide the finite-key analysis against coherent attacks and implement a proof-of-principle experiment to demonstrate our scheme's feasibility. Our scheme achieves a key rate of 85.3 bps under a 35 dB channel loss. Combined with security against general attacks and accessible technology, our protocol is a promising candidate for practical multiparty quantum communication networks.

quant-ph↗

arXiv2023

Source-independent quantum random number generator against tailored detector blinding attacks

Randomness, mainly in the form of random numbers, is the fundamental prerequisite for the security of many cryptographic tasks. Quantum randomness can be extracted even if adversaries are fully aware of the protocol and even control the randomness source. However, an adversary can further manipulate the randomness via tailored detector blinding attacks, which are hacking attacks suffered by protocols with trusted detectors. Here, by treating no-click events as valid events, we propose a quantum random number generation protocol that can simultaneously address source vulnerability and ferocious tailored detector blinding attacks. The method can be extended to high-dimensional random number generation. We experimentally demonstrate the ability of our protocol to generate random numbers for two-dimensional measurement with a generation speed of 0.1 bit per pulse.

quant-ph↗

arXiv2023

Complexity-Based Prompting for Multi-Step Reasoning

We study the task of prompting large-scale language models to perform multi-step reasoning. Existing work shows that when prompted with a chain of thoughts (CoT), sequences of short sentences describing intermediate reasoning steps towards a final answer, large language models can generate new reasoning chains and predict answers for new inputs. A central question is which reasoning examples make the most effective prompts. In this work, we propose complexity-based prompting, a simple and effective example selection scheme for multi-step reasoning. We show that prompts with higher reasoning complexity, i.e., chains with more reasoning steps, achieve substantially better performance on multi-step reasoning tasks over strong baselines. We further extend our complexity-based criteria from prompting (selecting inputs) to decoding (selecting outputs), where we sample multiple reasoning chains from the model, then choose the majority of generated answers from complex reasoning chains (over simple chains). When used to prompt GPT-3 and Codex, our approach substantially improves multi-step reasoning accuracy and achieves new state-of-the-art (SOTA) performance on three math benchmarks (GSM8K, MultiArith, and MathQA) and two BigBenchHard tasks (Date Understanding and Penguins), with an average +5.3 and up to +18 accuracy improvements. Compared with existing example selection schemes like manual tuning or retrieval-based selection, selection based on reasoning complexity is intuitive, easy to implement, and annotation-efficient. Further results demonstrate the robustness of performance gains from complex prompts under format perturbation and distribution shift.

cs.CL↗

arXiv2023

Specializing Smaller Language Models towards Multi-Step Reasoning

The surprising ability of Large Language Models (LLMs) to perform well on complex reasoning with only few-shot chain-of-thought prompts is believed to emerge only in very large-scale models (100+ billion parameters). We show that such abilities can, in fact, be distilled down from GPT-3.5 ($\ge$ 175B) to T5 variants ($\le$ 11B). We propose model specialization, to specialize the model's ability towards a target task. The hypothesis is that large models (commonly viewed as larger than 100B) have strong modeling power, but are spread on a large spectrum of tasks. Small models (commonly viewed as smaller than 10B) have limited model capacity, but if we concentrate their capacity on a specific target task, the model can achieve a decent improved performance. We use multi-step math reasoning as our testbed because it is a very typical emergent ability. We show two important aspects of model abilities: (1). there exists a very complex balance/ tradeoff between language models' multi-dimensional abilities; (2). by paying the price of decreased generic ability, we can clearly lift up the scaling curve of models smaller than 10B towards a specialized multi-step math reasoning ability. We further give comprehensive discussions about important design choices for better generalization, including the tuning data format, the start model checkpoint, and a new model selection method. We hope our practice and discoveries can serve as an important attempt towards specialized smaller models in the new research paradigm set by LLMs.

cs.CL↗

arXiv2022

All-Photonic Quantum Repeater for Multipartite Entanglement Generation

Quantum network applications like distributed quantum computing and quantum secret sharing present a promising future network equipped with quantum resources. Entanglement generation and distribution over long distances is critical and unavoidable to utilize quantum technology in a fully-connected network. The distribution of bipartite entanglement over long distances has seen some progresses, while the distribution of multipartite entanglement over long distances remains unsolved. Here we report a two-dimensional quantum repeater protocol for the generation of multipartite entanglement over long distances with all-photonic framework to fill this gap. The yield of the proposed protocol shows long transmission distance under various numbers of network users. With the improved efficiency and flexibility of extending the number of users, we anticipate that our protocol can work as a significant building block for quantum networks in the future.

quant-ph↗

arXiv2022

TorchOpt: An Efficient Library for Differentiable Optimization

Recent years have witnessed the booming of various differentiable optimization algorithms. These algorithms exhibit different execution patterns, and their execution needs massive computational resources that go beyond a single CPU and GPU. Existing differentiable optimization libraries, however, cannot support efficient algorithm development and multi-CPU/GPU execution, making the development of differentiable optimization algorithms often cumbersome and expensive. This paper introduces TorchOpt, a PyTorch-based efficient library for differentiable optimization. TorchOpt provides a unified and expressive differentiable optimization programming abstraction. This abstraction allows users to efficiently declare and analyze various differentiable optimization programs with explicit gradients, implicit gradients, and zero-order gradients. TorchOpt further provides a high-performance distributed execution runtime. This runtime can fully parallelize computation-intensive differentiation operations (e.g. tensor tree flattening) on CPUs / GPUs and automatically distribute computation to distributed devices. Experimental results show that TorchOpt achieves $5.2\times$ training time speedup on an 8-GPU server. TorchOpt is available at: https://github.com/metaopt/torchopt/.

cs.MS↗

arXiv2022

Just-DREAM-about-it: Figurative Language Understanding with DREAM-FLUTE

Figurative language (e.g., "he flew like the wind") is challenging to understand, as it is hard to tell what implicit information is being conveyed from the surface form alone. We hypothesize that to perform this task well, the reader needs to mentally elaborate the scene being described to identify a sensible meaning of the language. We present DREAM-FLUTE, a figurative language understanding system that does this, first forming a "mental model" of situations described in a premise and hypothesis before making an entailment/contradiction decision and generating an explanation. DREAM-FLUTE uses an existing scene elaboration model, DREAM, for constructing its "mental model." In the FigLang2022 Shared Task evaluation, DREAM-FLUTE achieved (joint) first place (Acc@60=63.3%), and can perform even better with ensemble techniques, demonstrating the effectiveness of this approach. More generally, this work suggests that adding a reflective component to pretrained language models can improve their performance beyond standard fine-tuning (3.3% improvement in Acc@60).

cs.CL↗

arXiv2022

Experimental measurement-device-independent type quantum key distribution with flawed and correlated sources

The security of quantum key distribution (QKD) is severely threatened by discrepancies between realistic devices and theoretical assumptions. Recently, a significant framework called the reference technique was proposed to provide security against arbitrary source flaws under current technology such as state preparation flaws, side channels caused by mode dependencies, the Trojan horse atttacks and pulse correlations. Here, we adopt the reference technique to prove security of an efficient four-phase measurement-device-independent QKD using laser pulses against potential source imperfections. We present a characterization of source flaws and connect them to experiments, together with a finite-key analysis against coherent attacks. In addition, we demonstrate the feasibility of our protocol through a proof-of-principle experimental implementation and achieve a secure key rate of 253 bps with a 20 dB channel loss. Compared with previous QKD protocols with imperfect devices, our study considerably improves both the secure key rate and the transmission distance, and shows application potential in the practical deployment of secure QKD with device imperfections.

quant-ph↗

arXiv2022

Measurement of the proton structure parameters in the forward-backward charge asymmetry

The forward-backward asymmetry ($A_{FB}$) in the Drell-Yan process $pp/p\bar p \to Z/γ^* \to \ell^+\ell^-$ is sensitive to the proton structure information. Such information has been factorized into well-defined proton structure parameters which can be regarded as experimental observables. In this paper, we extract the structure parameters from the $A_{FB}$ distributions reported by the CMS collaboration in $pp$ collisions at $\sqrt{s} = 8$ TeV, and by the D0 collaboration in $p\bar p$ collisions at $\sqrt{s} = 1.96$ TeV. It is the first time that the unique parton information in the $A_{FB}$ spectrum can be decoupled from the electroweak calculation and measured as standalone observables, which can be used as new data constraints in the global quantum chromodynamics analysis of the parton distribution functions (PDFs). Although the parton information in the $pp$ and $p\bar p$ collisions are different, and the precisions of the measured structure parameters are statistically limited, the results from both the hadron colliders indicate that the down quark contribution might be higher than the theoretical predictions with the current PDFs at the relevant momentum fraction range.

hep-ex↗

arXiv2022

Scaling Structured Inference with Randomization

Deep discrete structured models have seen considerable progress recently, but traditional inference using dynamic programming (DP) typically works with a small number of states (less than hundreds), which severely limits model capacity. At the same time, across machine learning, there is a recent trend of using randomized truncation techniques to accelerate computations involving large sums. Here, we propose a family of randomized dynamic programming (RDP) algorithms for scaling structured models to tens of thousands of latent states. Our method is widely applicable to classical DP-based inference (partition, marginal, reparameterization, entropy) and different graph structures (chains, trees, and more general hypergraphs). It is also compatible with automatic differentiation: it can be integrated with neural networks seamlessly and learned with gradient-based optimizers. Our core technique approximates the sum-product by restricting and reweighting DP on a small subset of nodes, which reduces computation by orders of magnitude. We further achieve low bias and variance via Rao-Blackwellization and importance sampling. Experiments over different graphs demonstrate the accuracy and efficiency of our approach. Furthermore, when using RDP for training a structured variational autoencoder with a scaled inference network, we achieve better test likelihood than baselines and successfully prevent posterior collapse. code at: https://github.com/FranxYao/RDP

cs.LG↗