Reinforcement Learning from Human Feedback (RLHF) is currently the most widely used method to align large language models (LLMs) with human preferences. Existing RLHF methods can be roughly categorized as either reward-based or reward-free. Novel applications such as ChatGPT and Claude leverage reward-based methods that first learn a reward model and apply actor-critic algorithms, such as Proximal Policy Optimization (PPO). However, in academic benchmarks, state-of-the-art results are often achieved via reward-free methods, such as Direct Preference Optimization (DPO). Is DPO truly superior to PPO? Why does PPO perform poorly on these benchmarks? In this paper, we first conduct both theoretical and empirical studies on the algorithmic properties of DPO and show that DPO may have fundamental limitations. Moreover, we also comprehensively examine PPO and reveal the key factors for the best performances of PPO in fine-tuning LLMs. Finally, we benchmark DPO and PPO across a collection of RLHF testbeds, ranging from dialogue to code generation. Experiment results demonstrate that PPO is able to surpass other alignment methods in all cases and achieve state-of-the-art results in challenging code competitions.

Researchers and developers increasingly rely on toxicity scoring to moderate generative language model outputs, in settings such as customer service, information retrieval, and content generation. However, toxicity scoring may render pertinent information inaccessible, rigidify or "value-lock" cultural norms, and prevent language reclamation processes, particularly for marginalized people. In this work, we extend the concept of algorithmic recourse to generative language models: we provide users a novel mechanism to achieve their desired prediction by dynamically setting thresholds for toxicity filtering. Users thereby exercise increased agency relative to interactions with the baseline system. A pilot study ($n = 30$) supports the potential of our proposed recourse mechanism, indicating improvements in usability compared to fixed-threshold toxicity-filtering of model outputs. Future work should explore the intersection of toxicity scoring, model controllability, user agency, and language reclamation processes -- particularly with regard to the bias that many communities encounter when interacting with generative language models.

This article presents Persistence Administered Collective Navigation (PACNav) as an approach for achieving decentralized collective navigation of Unmanned Aerial Vehicle (UAV) swarms. The technique is inspired by the flocking and collective navigation behavior observed in natural swarms, such as cattle herds, bird flocks, and even large groups of humans. PACNav relies solely on local observations of relative positions of UAVs, making it suitable for large swarms deprived of communication capabilities and external localization systems. We introduce the novel concepts of path persistence and path similarity, which allow each swarm member to analyze the motion of others. PACNav is grounded on two main principles: (1) UAVs with little variation in motion direction exhibit high path persistence and are considered reliable leaders by other UAVs; (2) groups of UAVs that move in a similar direction demonstrate high path similarity, and such groups are assumed to contain a reliable leader. The proposed approach also incorporates a reactive collision avoidance mechanism to prevent collisions with swarm members and environmental obstacles. The method is validated through simulated and real-world experiments conducted in a natural forest.

The vast majority of the popular English named entity recognition (NER) datasets contain American or British English data, despite the existence of many global varieties of English. As such, it is unclear whether they generalize for analyzing use of English globally. To test this, we build a newswire dataset, the Worldwide English NER Dataset, to analyze NER model performance on low-resource English variants from around the world. We test widely used NER toolkits and transformer models, including models using the pre-trained contextual models RoBERTa and ELECTRA, on three datasets: a commonly used British English newswire dataset, CoNLL 2003, a more American focused dataset OntoNotes, and our global dataset. All models trained on the CoNLL or OntoNotes datasets experienced significant performance drops-over 10 F1 in some cases-when tested on the Worldwide English dataset. Upon examination of region-specific errors, we observe the greatest performance drops for Oceania and Africa, while Asia and the Middle East had comparatively strong performance. Lastly, we find that a combined model trained on the Worldwide dataset and either CoNLL or OntoNotes lost only 1-2 F1 on both test sets.

Automatic Speech Recognition (ASR) can play a crucial role in enhancing the accessibility of spoken languages worldwide. In this paper, we build a set of ASR tools for Amharic, a language spoken by more than 50 million people primarily in eastern Africa. Amharic is written in the Ge'ez script, a sequence of graphemes with spacings denoting word boundaries. This makes computational processing of Amharic challenging since the location of spacings can significantly impact the meaning of formed sentences. We find that existing benchmarks for Amharic ASR do not account for these spacings and only measure individual grapheme error rates, leading to significantly inflated measurements of in-the-wild performance. In this paper, we first release corrected transcriptions of existing Amharic ASR test datasets, enabling the community to accurately evaluate progress. Furthermore, we introduce a post-processing approach using a transformer encoder-decoder architecture to organize raw ASR outputs into a grammatically complete and semantically meaningful Amharic sentence. Through experiments on the corrected test dataset, our model enhances the semantic correctness of Amharic speech recognition systems, achieving a Character Error Rate (CER) of 5.5\% and a Word Error Rate (WER) of 23.3\%.

For decades, Israel and Iran have fought a shadow war across the Middle East, trading attacks by land, sea, air and in cyberspace. A recent round of strikes -- mainly an aerial barrage by Iran against Israel last weekend -- has brought the conflict more clearly into the open and raised fears of a broader war. A retaliatory Israeli strike on an Iranian air base on Friday, however, appeared limited in scope, and analysts said it suggested an effort to pull back from the dangerous cycle and potentially move the war back into the shadows. August 2019: An Israeli airstrike killed two Iranian-trained militants in Syria, a drone set off a blast near a Hezbollah office in Lebanon and an airstrike in Qaim, Iraq, killed a commander of an Iran-backed Iraqi militia. Israel accused Iran at the time of trying to establish an overland arms-supply line through Iraq and northern Syria to Lebanon, and analysts said the strikes were aimed at stopping Iran and signaling to its proxies that Israel would not tolerate a fleet of smart missiles on its borders. January 2020: Israel greeted with satisfaction the assassination of Maj.

Israeli missiles have hit a site in Iran, ABC News reported late on Thursday, citing a U.S. official, days after Iran launched a drone strike on Israel in response to an attack at the Iranian embassy in Syria. Iran's Fars news agency said an explosion was heard at an airport in the Iranian city of Isafahan, but the cause was not immediately known. Several Iranian nuclear sites are located in Isfahan province, including Natanz, the centerpiece of Iran's uranium enrichment program. Several flights were diverted over Iranian airspace, CNN reported. Over the weekend, Iran launched hundreds of drones and missiles in a retaliatory strike after a suspected Israeli strike on its embassy compound in Syria.

We propose an approach for estimating the latent knowledge embedded inside large language models (LLMs). We leverage the in-context learning (ICL) abilities of LLMs to estimate the extent to which an LLM knows the facts stored in a knowledge base. Our knowledge estimator avoids reliability concerns with previous prompting-based methods, is both conceptually simpler and easier to apply, and we demonstrate that it can surface more of the latent knowledge embedded in LLMs. We also investigate how different design choices affect the performance of ICL-based knowledge estimation. Using the proposed estimator, we perform a large-scale evaluation of the factual knowledge of a variety of open source LLMs, like OPT, Pythia, Llama(2), Mistral, Gemma, etc. over a large set of relations and facts from the Wikidata knowledge base. We observe differences in the factual knowledge between different model families and models of different sizes, that some relations are consistently better known than others but that models differ in the precise facts they know, and differences in the knowledge of base models and their finetuned counterparts.

The safety and accuracy of robotic navigation hold paramount importance, especially in the realm of soft continuum robotics, where the limitations of traditional rigid sensors become evident. Encoders, piezoresistive, and potentiometer sensors often fail to integrate well with the flexible nature of these robots, adding unwanted bulk and rigidity. To overcome these hurdles, our study presents a new approach to shape sensing in soft continuum robots through the use of soft e-textile resistive sensors. This sensor, designed to flawlessly integrate with the robot's structure, utilizes a resistive material that adjusts its resistance in response to the robot's movements and deformations. This adjustment facilitates the capture of multidimensional force measurements across the soft sensor layers. A deep Convolutional Neural Network (CNN) is employed to decode the sensor signals, enabling precise estimation of the robot's shape configuration based on the detailed data from the e-textile sensor. Our research investigates the efficacy of this e-textile sensor in determining the curvature parameters of soft continuum robots. The findings are encouraging, showing that the soft e-textile sensor not only matches but potentially exceeds the capabilities of traditional rigid sensors in terms of shape sensing and estimation. This advancement significantly boosts the safety and efficiency of robotic navigation systems.

In recent years, Vision Transformers (ViTs) have shown promising classification performance over Convolutional Neural Networks (CNNs) due to their self-attention mechanism. Many researchers have incorporated ViTs for Hyperspectral Image (HSI) classification. HSIs are characterised by narrow contiguous spectral bands, providing rich spectral data. Although ViTs excel with sequential data, they cannot extract spectral-spatial information like CNNs. Furthermore, to have high classification performance, there should be a strong interaction between the HSI token and the class (CLS) token. To solve these issues, we propose a 3D-Convolution guided Spectral-Spatial Transformer (3D-ConvSST) for HSI classification that utilizes a 3D-Convolution Guided Residual Module (CGRM) in-between encoders to "fuse" the local spatial and spectral information and to enhance the feature propagation. Furthermore, we forego the class token and instead apply Global Average Pooling, which effectively encodes more discriminative and pertinent high-level features for classification. Extensive experiments have been conducted on three public HSI datasets to show the superiority of the proposed model over state-of-the-art traditional, convolutional, and Transformer models. The code is available at https://github.com/ShyamVarahagiri/3D-ConvSST.