Code generation models have increasingly become integral to aiding software development, offering assistance in tasks such as code completion, debugging, and code translation. Although current research has thoroughly examined the correctness of code produced by code generation models, a vital aspect, i.e., the efficiency of the generated code, has often been neglected. This paper presents EffiBench, a benchmark with 1,000 efficiency-critical coding problems for assessing the efficiency of code generated by code generation models. EffiBench contains a diverse set of LeetCode coding problems. Each problem is paired with an executable human-written canonical solution. With EffiBench, we empirically examine the capability of 21 Large Language Models (13 open-sourced and 8 closed-sourced) in generating efficient code. The results demonstrate that GPT-4-turbo generates the most efficient code, significantly outperforming Palm-2-chat-bison, Claude-instant-1, Gemini-pro, GPT-4, and GPT-3.5. Nevertheless, its code efficiency is still worse than the efficiency of human-written canonical solutions. In particular, the average and worst execution time of GPT-4-turbo generated code is 1.69 and 45.49 times that of the canonical solutions.

Current methods for large language model alignment typically use scalar human preference labels. However, this convention tends to oversimplify the multi-dimensional and heterogeneous nature of human preferences, leading to reduced expressivity and even misalignment. This paper presents Panacea, an innovative approach that reframes alignment as a multi-dimensional preference optimization problem. Panacea trains a single model capable of adapting online and Pareto-optimally to diverse sets of preferences without the need for further tuning. A major challenge here is using a low-dimensional preference vector to guide the model's behavior, despite it being governed by an overwhelmingly large number of parameters. To address this, Panacea is designed to use singular value decomposition (SVD)-based low-rank adaptation, which allows the preference vector to be simply injected online as singular values. Theoretically, we prove that Panacea recovers the entire Pareto front with common loss aggregation methods under mild conditions. Moreover, our experiments demonstrate, for the first time, the feasibility of aligning a single LLM to represent a spectrum of human preferences through various optimization methods. Our work marks a step forward in effectively and efficiently aligning models to diverse and intricate human preferences in a controllable and Pareto-optimal manner.

Existing methods for evaluating large language models face challenges such as data contamination, sensitivity to prompts, and the high cost of benchmark creation. To address this, we propose a lossless data compression based evaluation approach that tests how models' predictive abilities generalize after their training cutoff. Specifically, we collect comprehensive test data spanning 83 months from 2017 to 2023 and split the data into training and testing periods according to models' training data cutoff. We measure: 1) the compression performance on the testing period as a measure of generalization on unseen data; and 2) the performance gap between the training and testing period as a measure of robustness. Our experiments test 14 representative large language models with various sizes on sources including Wikipedia, news articles, code, arXiv papers, and multi-modal data. We find that the compression rate of many models reduces significantly after their cutoff date, but models such as Mistral and Llama-2 demonstrate a good balance between performance and robustness. Results also suggest that models struggle to generalize on news and code data, but work especially well on arXiv papers. We also find the context size and tokenization implementation have a big impact of on the overall compression performance.

Large language models (LLMs) such as ChatGPT have garnered substantial attention in the media and among the machine learning (ML) community. LLMs represent a pivotal paradigm shift in artificial intelligence (AI), consisting of transformer architectures substantially larger in scale compared to their predecessors, such as Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks [1], and leverage internet-scale text corpora, thus excelling not only on text completion tasks but demonstrating emergent capabilities in rudimentary language reasoning [2, 3]. LLMs display several features conducive to the small data regime present in most EEG datasets, where the largest datasets typically have on the order of only thousands of EEGs. Primarily, LLMs have the capability to perform fewand even zero-shot learning [4]. Recent research has investigated how LLMs can perform few-shot learning in domains ranging from cancer drug synergy prediction to cardiac signal analysis [5, 6]. Other work has demonstrated the ability of LLMs to outperform experts in annotating political Twitter messages with zero-shot learning [7]. Additionally, previous work has shown that transformer architectures are capable of utilizing in-context learning for zero-shot tasks - in other words, utilizing information provided in the prompt in order to yield better performance on various tasks [8].

The training of large language models (LLMs) necessitates substantial data and computational resources, and updating outdated LLMs entails significant efforts and resources. While numerous model editing techniques (METs) have emerged to efficiently update model outputs without retraining, their effectiveness in multilingual LLMs, where knowledge is stored in diverse languages, remains an underexplored research area. This research paper introduces the cross-lingual model editing (\textbf{XME}) paradigm, wherein a fact is edited in one language, and the subsequent update propagation is observed across other languages. To investigate the XME paradigm, we conducted experiments using BLOOM, mBERT, and XLM-RoBERTa using the two writing scripts: \textit{Latin} (English, French, and Spanish) and \textit{Indic} (Hindi, Gujarati, and Bengali). The results reveal notable performance limitations of state-of-the-art METs under the XME setting, mainly when the languages involved belong to two distinct script families. These findings highlight the need for further research and development of XME techniques to address these challenges. For more comprehensive information, the dataset used in this research and the associated code are publicly available at the following URL\url{https://github.com/lingo-iitgn/XME}.

Large Language Models (LLMs) face limitations due to the high demand on GPU memory and computational resources when handling long contexts. While sparsify the Key-Value (KV) cache of transformer model is a typical strategy to alleviate resource usage, it unavoidably results in the loss of information. We introduce Lossless Compressed Memory Attention (LoMA), a novel approach that enables lossless compression of the KV cache, thereby reducing the memory and computational demands during autoregressive generation. LoMA incorporates a specialized training or fine-tuning precedure alongside an autoregressive generation algorithm optimized for the compressed context. Our method compresses the KV cache after every $tc$ generated tokens with a compression ratio of $c$ and a target compressed length $t$, and this process occurs within a single inference pass without dependency on auxiliary models. We engineered an efficient training scheme involving specific inputs, attention masks, and position identifiers to instill this compression capability. Experimental validation has demonstrated that LoMA significantly reducing computational consumption and memory usage through achieving lossless KV cache compression.

Recent advancements in large language models (LLMs) have provided a new avenue for chatbot development, while most existing research has primarily centered on single-user chatbots that focus on deciding "What" to answer after user inputs. In this paper, we identified that multi-user chatbots have more complex 3W design dimensions -- "What" to say, "When" to respond, and "Who" to answer. Additionally, we proposed Multi-User Chat Assistant (MUCA), which is an LLM-based framework for chatbots specifically designed for group discussions. MUCA consists of three main modules: Sub-topic Generator, Dialog Analyzer, and Utterance Strategies Arbitrator. These modules jointly determine suitable response contents, timings, and the appropriate recipients. To make the optimizing process for MUCA easier, we further propose an LLM-based Multi-User Simulator (MUS) that can mimic real user behavior. This enables faster simulation of a conversation between the chatbot and simulated users, making the early development of the chatbot framework much more efficient. MUCA demonstrates effectiveness, including appropriate chime-in timing, relevant content, and positive user engagement, in goal-oriented conversations with a small to medium number of participants, as evidenced by case studies and experimental results from user studies.

It is well known that LLMs cannot generalize well to long contexts whose lengths are larger than the training sequence length. This poses challenges when employing LLMs for processing long input sequences during inference. In this work, we argue that LLMs themselves have inherent capabilities to handle long contexts without fine-tuning. To achieve this goal, we propose SelfExtend to extend the context window of LLMs by constructing bi-level attention information: the grouped attention and the neighbor attention. The grouped attention captures the dependencies among tokens that are far apart, while neighbor attention captures dependencies among adjacent tokens within a specified range. The two-level attentions are computed based on the original model's self-attention mechanism during inference. With minor code modification, our SelfExtend can effortlessly extend existing LLMs' context window without any fine-tuning. We conduct comprehensive experiments on multiple benchmarks and the results show that our SelfExtend can effectively extend existing LLMs' context window length. The code can be found at \url{https://github.com/datamllab/LongLM}.

The development of large language models (LLMs) has been catalyzed by advancements in pre-training techniques. These models have demonstrated robust reasoning capabilities through manually designed prompts. In this work, we evaluate the conversational reasoning capabilities of the current state-of-the-art LLM (GPT-4) on knowledge graphs (KGs). However, the performance of LLMs is constrained due to a lack of KG environment awareness and the difficulties in developing effective optimization mechanisms for intermediary reasoning stages. We further introduce LLM-ARK, a LLM grounded KG reasoning agent designed to deliver precise and adaptable predictions on KG paths. LLM-ARK leverages Full Textual Environment (FTE) prompt to assimilate state information within each reasoning step. We reframe the challenge of multi-hop reasoning on the KG as a sequential decision-making task. Utilizing the Proximal Policy Optimization (PPO) online policy gradient reinforcement learning algorithm, our model is optimized to learn from rich reward signals. Additionally, we conduct an evaluation of our model and GPT-4 on the OpenDialKG dataset. The experimental results reveal that LLaMA-2-7B-ARK outperforms the current state-of-the-art model by 5.28 percentage points, with a performance rate of 36.39% on the target@1 evaluation metric. Meanwhile, GPT-4 scored 14.91%, further demonstrating the effectiveness of our method. Our code is available on GitHub (https://github.com/Aipura/LLM-ARK) for further access.

Language-conditioned robotic manipulation represents a cutting-edge area of research, enabling seamless communication and cooperation between humans and robotic agents. This field focuses on teaching robotic systems to comprehend and execute instructions conveyed in natural language. To achieve this, the development of robust language understanding models capable of extracting actionable insights from textual input is essential. In this comprehensive survey, we systematically explore recent advancements in language-conditioned approaches within the context of robotic manipulation. We analyze these approaches based on their learning paradigms, which encompass reinforcement learning, imitation learning, and the integration of foundational models, such as large language models and vision-language models. Furthermore, we conduct an in-depth comparative analysis, considering aspects like semantic information extraction, environment & evaluation, auxiliary tasks, and task representation. Finally, we outline potential future research directions in the realm of language-conditioned learning for robotic manipulation, with the topic of generalization capabilities and safety issues. The GitHub repository of this paper can be found at https://github.com/hk-zh/language-conditioned-robot-manipulation-models