In recent years, studies have been actively conducted on combining large language models (LLM) and robotics; however, most have not considered end-to-end feedback in the robot-motion generation phase. The prediction of deep neural networks must contain errors, it is required to update the trained model to correspond to the real environment to generate robot motion adaptively. This study proposes an integration method that connects the robot-motion learning model and LLM using shared latent variables. When generating robot motion, the proposed method updates shared parameters based on prediction errors from both sensorimotor attention points and task language instructions given to the robot. This allows the model to search for latent parameters appropriate for the robot task efficiently. Through simulator experiments on multiple robot tasks, we demonstrated the effectiveness of our proposed method from two perspectives: position generalization and language instruction generalization abilities.

Seeking answers to questions within long scientific research articles is a crucial area of study that aids readers in quickly addressing their inquiries. However, existing question-answering (QA) datasets based on scientific papers are limited in scale and focus solely on textual content. To address this limitation, we introduce SPIQA (Scientific Paper Image Question Answering), the first large-scale QA dataset specifically designed to interpret complex figures and tables within the context of scientific research articles across various domains of computer science. Leveraging the breadth of expertise and ability of multimodal large language models (MLLMs) to understand figures, we employ automatic and manual curation to create the dataset. We craft an information-seeking task involving multiple images that cover a wide variety of plots, charts, tables, schematic diagrams, and result visualizations. SPIQA comprises 270K questions divided into training, validation, and three different evaluation splits. Through extensive experiments with 12 prominent foundational models, we evaluate the ability of current multimodal systems to comprehend the nuanced aspects of research articles. Additionally, we propose a Chain-of-Thought (CoT) evaluation strategy with in-context retrieval that allows fine-grained, step-by-step assessment and improves model performance. We further explore the upper bounds of performance enhancement with additional textual information, highlighting its promising potential for future research and the dataset's impact on revolutionizing how we interact with scientific literature.

Contemporary language models are increasingly multilingual, but Chinese LLM developers must navigate complex political and business considerations of language diversity. Language policy in China aims at influencing the public discourse and governing a multi-ethnic society, and has gradually transitioned from a pluralist to a more assimilationist approach since 1949. We explore the impact of these influences on current language technology. We evaluate six open-source multilingual LLMs pre-trained by Chinese companies on 18 languages, spanning a wide range of Chinese, Asian, and Anglo-European languages. Our experiments show Chinese LLMs performance on diverse languages is indistinguishable from international LLMs. Similarly, the models' technical reports also show lack of consideration for pretraining data language coverage except for English and Mandarin Chinese. Examining Chinese AI policy, model experiments, and technical reports, we find no sign of any consistent policy, either for or against, language diversity in China's LLM development. This leaves a puzzling fact that while China regulates both the languages people use daily as well as language model development, they do not seem to have any policy on the languages in language models.

Hallucination detection in Large Language Models (LLMs) is crucial for ensuring their reliability. This work presents our participation in the CLEF ELOQUENT HalluciGen shared task, where the goal is to develop evaluators for both generating and detecting hallucinated content. We explored the capabilities of four LLMs: Llama 3, Gemma, GPT-3.5 Turbo, and GPT-4, for this purpose. We also employed ensemble majority voting to incorporate all four models for the detection task. The results provide valuable insights into the strengths and weaknesses of these LLMs in handling hallucination generation and detection tasks.

For the Transformer architecture [59], the attention mechanism constitutes the primary computational bottleneck, since computing the self-attention scores of queries and keys has quadratic scaling in the sequence length. Scaling attention to longer context will unlock new capabilities (modeling and reasoning over multiple long documents [24, 43, 50] and files in large codebases [30, 48]), new modalities (high-resolution images [11], audio [23], video [25]), and new applications (user interaction with long history [53], agent workflow with long horizon [62]). This has generated significant interest in making attention faster in the long-context regime, including by approximation [14, 27, 56] and software optimization ([17, 29, 45]), or even alternative architectures [22, 42, 55]. In this work, we build on the work of Dao et al. [17] on developing exact-attention algorithms that integrate knowledge of the GPU's execution model and hardware characteristics into their high-level design. In [17], Dao et al. introduced FlashAttention, a novel tiling strategy for parallelizing attention that eliminates intermediate reads/writes to slow global memory through fusing all of the attention operations into a single GPU kernel. Dao [15] restructured the algorithm as FlashAttention-2 to also parallelize over the sequence length dimension and perform the inner loop of the forward pass over blocks of the key and value matrices, thus improving the occupancy and distribution of work on the GPU. However, we observe that FlashAttention-2 nonetheless achieves poor utilization on newer GPUs relative to optimized matrix-multiplication (GEMM) kernels, such as 35% vs. 80-90% on the Hopper H100 GPU. Partially, this may be attributed to implementation-level differences, such as not using Hopper-specific instructions in place of Ampere ones when targeting the Tensor Cores. Several work such as ThunkerKitten [52] and cuDNN 9 [39] has shown that with Hopper-specific instructions and tile-based abstractions, one can speedup attention computation and simplify the implementation.

Despite their nearly universal adoption for large language models, the internal workings of transformers are not well understood. We aim to better understand the impact of removing or reorganizing information throughout the layers of a pretrained transformer. Such an understanding could both yield better usage of existing models as well as to make architectural improvements to produce new variants. We present a series of empirical studies on frozen models that show that the lower and final layers of pretrained transformers differ from middle layers, but that middle layers have a surprising amount of uniformity. We further show that some classes of problems have robustness to skipping layers, running the layers in an order different from how they were trained, or running the layers in parallel. Our observations suggest that even frozen pretrained models may gracefully trade accuracy for latency by skipping layers or running layers in parallel.

Typical schemes for automated red-teaming large language models (LLMs) focus on discovering prompts that trigger a frozen language model (the defender) to generate toxic text. This often results in the prompting model (the adversary) producing text that is unintelligible and unlikely to arise. Here, we propose a reinforcement learning formulation of the LLM red-teaming task which allows us to discover prompts that both (1) trigger toxic outputs from a frozen defender and (2) have low perplexity as scored by the defender. We argue these cases are most pertinent in a red-teaming setting because of their likelihood to arise during normal use of the defender model. We solve this formulation through a novel online and weakly supervised variant of Identity Preference Optimization (IPO) on GPT-2 and GPT-2 XL defenders. We demonstrate that our policy is capable of generating likely prompts that also trigger toxicity. Finally, we qualitatively analyze learned strategies, trade-offs of likelihood and toxicity, and discuss implications. Source code is available for this project at: https://github.com/sisl/ASTPrompter/.

Large Language Models (LLMs) have demonstrated impressive problem-solving capabilities in mathematics through step-by-step reasoning chains. However, they are susceptible to reasoning errors that impact the quality of subsequent reasoning chains and the final answer due to language models' autoregressive token-by-token generating nature. Recent works have proposed adopting external verifiers to guide the generation of reasoning paths, but existing works utilize models that have been trained with step-by-step labels to assess the correctness of token-by-token reasoning chains. Consequently, they struggle to recognize discriminative details of tokens within a reasoning path and lack the ability to evaluate whether an intermediate reasoning path is on a promising track toward the correct final answer. To amend the lack of sound and token-grained math-verification signals, we devise a novel training scheme for verifiers that apply token-level supervision with the expected cumulative reward (i.e., value). Furthermore, we propose a practical formulation of the cumulative reward by reducing it to finding the probability of future correctness of the final answer and thereby enabling the empirical estimation of the value. Experimental results on mathematical reasoning benchmarks show that Token-Supervised Value Model (TVM) can outperform step-by-step verifiers on GSM8K and MATH with Mistral and Llama.

Object-oriented embodied navigation aims to locate specific objects, defined by category or depicted in images. Existing methods often struggle to generalize to open vocabulary goals without extensive training data. While recent advances in Vision-Language Models (VLMs) offer a promising solution by extending object recognition beyond predefined categories, efficient goal-oriented exploration becomes more challenging in an open vocabulary setting. We introduce OVExp, a learning-based framework that integrates VLMs for Open-Vocabulary Exploration. OVExp constructs scene representations by encoding observations with VLMs and projecting them onto top-down maps for goal-conditioned exploration. Goals are encoded in the same VLM feature space, and a lightweight transformer-based decoder predicts target locations while maintaining versatile representation abilities. To address the impracticality of fusing dense pixel embeddings with full 3D scene reconstruction for training, we propose constructing maps using low-cost semantic categories and transforming them into CLIP's embedding space via the text encoder. The simple but effective design of OVExp significantly reduces computational costs and demonstrates strong generalization abilities to various navigation settings. Experiments on established benchmarks show OVExp outperforms previous zero-shot methods, can generalize to diverse scenes, and handle different goal modalities.

Retrieval-Augmented Generation (RAG) allows overcoming the limited knowledge of LLMs by extending the input with external information. As a consequence, the contextual inputs to the model become much longer which slows down decoding time directly translating to the time a user has to wait for an answer. We address this challenge by presenting COCOM, an effective context compression method, reducing long contexts to only a handful of Context Embeddings speeding up the generation time by a large margin. Our method allows for different compression rates trading off decoding time for answer quality. Compared to earlier methods, COCOM allows for handling multiple contexts more effectively, significantly reducing decoding time for long inputs. Our method demonstrates a speed-up of up to 5.69 $\times$ while achieving higher performance compared to existing efficient context compression methods.