Amidst the recent strides in evaluating Large Language Models for Code (Code-LLMs), existing benchmarks have mainly focused on functional correctness, overlooking the importance of computational efficiency. To fill the gap, we present Mercury, the first computational efficiency benchmark for Code-LLMs. It comprises 1,889 Python tasks, each with adequate solutions to support a runtime distribution. Based on the distribution, we introduce a new metric Beyond, which computes a runtime-percentile-weighted Pass score to reflect functional correctness and computational efficiency simultaneously. On Mercury, leading Code-LLMs can achieve 67% on Pass, while less than 50% on Beyond. Given that an ideal Beyond score would be aligned with the Pass score, it indicates that while Code-LLMs exhibit impressive capabilities in generating functionally correct code, there remains a notable gap in their efficiency. Finally, our empirical experiments reveal that Direct Preference Optimization (DPO) serves as a robust baseline for enhancing computational efficiency compared with Supervised Fine Tuning (SFT), which paves a promising avenue for future exploration of efficient code generation.
相關內容
With the advancement of deep learning techniques, the performance of Automatic Program Repair(APR) techniques has reached a new level. Previous deep learning-based APR techniques essentially modified program sentences in the Autoregressive(AR) manner, which predicts future values based on past values. Due to the manner of word-by-word generation, the AR-based APR technique has a huge time delay. This negative consequence overshadows the widespread adoption of APR techniques in real-life software development. To address the issue, we aim to apply the Non-Autoregressive(NAR) method to the APR task, which can output target code in a parallel manner to avoid huge inference delays. To effectively adapt the NAR manner for the APR task, we in this paper propose NARRepair, the first customized NAR code generation model for the APR task. The NARRepair features three major novelties, including 1) using repair actions to alleviate the over-correction issue, 2) extracting dependency information from AST to alleviate the issue of lacking inter-word dependency information, 3) employing two-stage decoding to alleviate the issue of lacking contextual information. We evaluated NARRepair on three widely used datasets in the APR community, and the results show that our technique can significantly improve the inference speed while maintaining high repair accuracy.
Recently, there have been significant advancements in Image Restoration based on CNN and transformer. However, the inherent characteristics of the Image Restoration task are often overlooked in many works. They, instead, tend to focus on the basic block design and stack numerous such blocks to the model, leading to parameters redundant and computations unnecessary. Thus, the efficiency of the image restoration is hindered. In this paper, we propose a Lightweight Baseline network for Image Restoration called LIR to efficiently restore the image and remove degradations. First of all, through an ingenious structural design, LIR removes the degradations existing in the local and global residual connections that are ignored by modern networks. Then, a Lightweight Adaptive Attention (LAA) Block is introduced which is mainly composed of proposed Adaptive Filters and Attention Blocks. The proposed Adaptive Filter is used to adaptively extract high-frequency information and enhance object contours in various IR tasks, and Attention Block involves a novel Patch Attention module to approximate the self-attention part of the transformer. On the deraining task, our LIR achieves the state-of-the-art Structure Similarity Index Measure (SSIM) and comparable performance to state-of-the-art models on Peak Signal-to-Noise Ratio (PSNR). For denoising, dehazing, and deblurring tasks, LIR also achieves a comparable performance to state-of-the-art models with a parameter size of about 30\%. In addition, it is worth noting that our LIR produces better visual results that are more in line with the human aesthetic.
This paper introduces DiffTOP, which utilizes Differentiable Trajectory OPtimization as the policy representation to generate actions for deep reinforcement and imitation learning. Trajectory optimization is a powerful and widely used algorithm in control, parameterized by a cost and a dynamics function. The key to our approach is to leverage the recent progress in differentiable trajectory optimization, which enables computing the gradients of the loss with respect to the parameters of trajectory optimization. As a result, the cost and dynamics functions of trajectory optimization can be learned end-to-end. DiffTOP addresses the ``objective mismatch'' issue of prior model-based RL algorithms, as the dynamics model in DiffTOP is learned to directly maximize task performance by differentiating the policy gradient loss through the trajectory optimization process. We further benchmark DiffTOP for imitation learning on standard robotic manipulation task suites with high-dimensional sensory observations and compare our method to feed-forward policy classes as well as Energy-Based Models (EBM) and Diffusion. Across 15 model-based RL tasks and 35imitation learning tasks with high-dimensional image and point cloud inputs, DiffTOP outperforms prior state-of-the-art methods in both domains.
Previous studies have shown that demonstrations can significantly help Large Language Models (LLMs ) perform better on the given tasks. However, this so-called In-Context Learning ( ICL ) ability is very sensitive to the presenting context, and often dozens of demonstrations are needed. In this work, we investigate if we can reduce the shot number while still maintaining a competitive performance. We present SeCoKD, a self-Knowledge Distillation ( KD ) training framework that aligns the student model with a heavily prompted variation, thereby increasing the utilization of a single demonstration. We experiment with the SeCoKD across three LLMs and six benchmarks focusing mainly on reasoning tasks. Results show that our method outperforms the base model and Supervised Fine-tuning ( SFT ), especially in zero-shot and one-shot settings by 30% and 10%, respectively. Moreover, SeCoKD brings little negative artifacts when evaluated on new tasks, which is more robust than Supervised Fine-tuning.
Reasoning about time is essential for Large Language Models (LLMs) to understand the world. Previous works focus on solving specific tasks, primarily on time-sensitive question answering. While these methods have proven effective, they cannot generalize to a wider spectrum of temporal reasoning tasks. Therefore, we propose a crucial question: Can we build a universal framework to handle a variety of temporal reasoning tasks? To that end, we systematically study 38 temporal reasoning tasks. Based on the observation that 19 tasks are directly related to mathematics, we first leverage the available mathematical dataset to set a solid foundation for temporal reasoning. However, the in-depth study indicates that focusing solely on mathematical enhancement falls short of addressing pure temporal reasoning tasks. To mitigate this limitation, we propose a simple but effective self-critic temporal optimization method to enhance the model's temporal reasoning capabilities without sacrificing general task abilities. Finally, we develop Timo, a model designed to excel in temporal reasoning at the 7B and 13B scales. Notably, Timo outperforms the counterpart LLMs by 10.0 and 7.6 in average accuracy scores and achieves the new state-of-the-art (SOTA) performance of comparable size. Extensive experiments further validate our framework's effectiveness and its generalization across diverse temporal tasks. The code is available at //github.com/zhaochen0110/Timo.
Machine Learning (ML) for Mineral Prospectivity Mapping (MPM) remains a challenging problem as it requires the analysis of associations between large-scale multi-modal geospatial data and few historical mineral commodity observations (positive labels). Recent MPM works have explored Deep Learning (DL) as a modeling tool with more representation capacity. However, these overparameterized methods may be more prone to overfitting due to their reliance on scarce labeled data. While a large quantity of unlabeled geospatial data exists, no prior MPM works have considered using such information in a self-supervised manner. Our MPM approach uses a masked image modeling framework to pretrain a backbone neural network in a self-supervised manner using unlabeled geospatial data alone. After pretraining, the backbone network provides feature extraction for downstream MPM tasks. We evaluated our approach alongside existing methods to assess mineral prospectivity of Mississippi Valley Type (MVT) and Clastic-Dominated (CD) Lead-Zinc deposits in North America and Australia. Our results demonstrate that self-supervision promotes robustness in learned features, improving prospectivity predictions. Additionally, we leverage explainable artificial intelligence techniques to demonstrate that individual predictions can be interpreted from a geological perspective.
As part of an ongoing worldwide effort to comprehend and monitor insect biodiversity, this paper presents the BIOSCAN-5M Insect dataset to the machine learning community and establish several benchmark tasks. BIOSCAN-5M is a comprehensive dataset containing multi-modal information for over 5 million insect specimens, and it significantly expands existing image-based biological datasets by including taxonomic labels, raw nucleotide barcode sequences, assigned barcode index numbers, and geographical information. We propose three benchmark experiments to demonstrate the impact of the multi-modal data types on the classification and clustering accuracy. First, we pretrain a masked language model on the DNA barcode sequences of the \mbox{BIOSCAN-5M} dataset, and demonstrate the impact of using this large reference library on species- and genus-level classification performance. Second, we propose a zero-shot transfer learning task applied to images and DNA barcodes to cluster feature embeddings obtained from self-supervised learning, to investigate whether meaningful clusters can be derived from these representation embeddings. Third, we benchmark multi-modality by performing contrastive learning on DNA barcodes, image data, and taxonomic information. This yields a general shared embedding space enabling taxonomic classification using multiple types of information and modalities. The code repository of the BIOSCAN-5M Insect dataset is available at {\url{//github.com/zahrag/BIOSCAN-5M}}
Object detectors usually achieve promising results with the supervision of complete instance annotations. However, their performance is far from satisfactory with sparse instance annotations. Most existing methods for sparsely annotated object detection either re-weight the loss of hard negative samples or convert the unlabeled instances into ignored regions to reduce the interference of false negatives. We argue that these strategies are insufficient since they can at most alleviate the negative effect caused by missing annotations. In this paper, we propose a simple but effective mechanism, called Co-mining, for sparsely annotated object detection. In our Co-mining, two branches of a Siamese network predict the pseudo-label sets for each other. To enhance multi-view learning and better mine unlabeled instances, the original image and corresponding augmented image are used as the inputs of two branches of the Siamese network, respectively. Co-mining can serve as a general training mechanism applied to most of modern object detectors. Experiments are performed on MS COCO dataset with three different sparsely annotated settings using two typical frameworks: anchor-based detector RetinaNet and anchor-free detector FCOS. Experimental results show that our Co-mining with RetinaNet achieves 1.4%~2.1% improvements compared with different baselines and surpasses existing methods under the same sparsely annotated setting.
We present CoDEx, a set of knowledge graph completion datasets extracted from Wikidata and Wikipedia that improve upon existing knowledge graph completion benchmarks in scope and level of difficulty. In terms of scope, CoDEx comprises three knowledge graphs varying in size and structure, multilingual descriptions of entities and relations, and tens of thousands of hard negative triples that are plausible but verified to be false. To characterize CoDEx, we contribute thorough empirical analyses and benchmarking experiments. First, we analyze each CoDEx dataset in terms of logical relation patterns. Next, we report baseline link prediction and triple classification results on CoDEx for five extensively tuned embedding models. Finally, we differentiate CoDEx from the popular FB15K-237 knowledge graph completion dataset by showing that CoDEx covers more diverse and interpretable content, and is a more difficult link prediction benchmark. Data, code, and pretrained models are available at //bit.ly/2EPbrJs.
ASR (automatic speech recognition) systems like Siri, Alexa, Google Voice or Cortana has become quite popular recently. One of the key techniques enabling the practical use of such systems in people's daily life is deep learning. Though deep learning in computer vision is known to be vulnerable to adversarial perturbations, little is known whether such perturbations are still valid on the practical speech recognition. In this paper, we not only demonstrate such attacks can happen in reality, but also show that the attacks can be systematically conducted. To minimize users' attention, we choose to embed the voice commands into a song, called CommandSong. In this way, the song carrying the command can spread through radio, TV or even any media player installed in the portable devices like smartphones, potentially impacting millions of users in long distance. In particular, we overcome two major challenges: minimizing the revision of a song in the process of embedding commands, and letting the CommandSong spread through the air without losing the voice "command". Our evaluation demonstrates that we can craft random songs to "carry" any commands and the modify is extremely difficult to be noticed. Specially, the physical attack that we play the CommandSongs over the air and record them can success with 94 percentage.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191